Post on 20-Nov-2021
Optimization of Material and Energy
Integration in Eco-Industrial
Networks
by
Ivan Kantor
A thesis
presented to the University of Waterloo
in fulfillment of the
thesis requirement for the degree of
Doctor of Philosophy
in
Chemical Engineering
Waterloo, Ontario, Canada, 2014
c© Ivan Kantor 2014
This thesis consists of material all of which I authored or co-authored: see Statement
of Contributions included in the thesis. This is a true copy of the thesis, including any
required final revisions, as accepted by my examiners.
I understand that my thesis may be made electronically available to the public.
ii
Chapter 3 is based on the previously published work “Air quality and environmental
impacts of alternative vehicle technologies in Ontario, Canada” by Kantor et al. [41] as seen
in the International Journal of Hydrogen Energy 35(10):5145-5153 and is reproduced with
permission from the International Association of Hydrogen Energy. The thesis author’s
specific contributions to this paper were to develop the model of emission reduction poten-
tials, conduct the simulations, prepare the graphics and results, write the final manuscript
and respond to the comments of reviewers. This work was conducted with direction from
the project supervisors, Dr. M. Fowler and Dr. A Elkamel, who are co-authors on the pub-
lication. Amirhossein Hajimiragha contributed with primary modeling of the electricity
grid in the province of Ontario to determine the supportable penetration of alternative-fuel
vehicles. This effort is reflected in the Appendix.
Chapter 4 is based on previously published work “Optimized production of hydrogen
in an eco-park network accounting for life-cycle emissions and profit” by Kantor et al.
[69] as seen in the International Journal of Hydrogen Energy 37(6):5347-5359 and is re-
produced with permission from the International Association of Hydrogen Energy. The
thesis author’s specific contributions to this paper were to develop the model, conduct
the simulations, prepare the graphics and results, write the final manuscript and respond
to the comments of reviewers. This work was conducted with direction from the project
supervisors, Dr. M.W. Fowler and Dr. A. Elkamel, who are co-authors on the publication.
Chapter 5 is based on the forthcoming work “Optimization of Material and Energy
Exchange in an Eco-park Network Considering Three Fuel Sources” by Kantor et al. [108],
in press with the International Journal of Advanced Operations Management and is repro-
duced with permission from the International Journal of Advanced Operations Manage-
ment. The thesis author’s specific contributions to this paper were to develop the model,
conduct the simulations, prepare the graphics and results, write the final manuscript and
iii
respond to the comments of reviewers. This work was conducted with direction from the
project supervisors, Dr. M.W. Fowler and Dr. A. Elkamel, who are co-authors on the
publication.
Chapter 6 is based on work submitted to the Journal of Cleaner Production entitled
“Generalized MINLP Modeling of Eco-Industrial Networks”. The thesis author’s specific
contributions to this paper were to develop the model, conduct the simulations, prepare
the graphics and results, write the final manuscript and submit to the journal with an
expectation of also responding to the comments of reviewers. Alberto Betancourt assisted
with the implementation of the model in GAMS and also assisted with the modeling doc-
umentation. This work was conducted with direction from the project supervisors, Dr.
M.W. Fowler and Dr. A. Elkamel, who are co-authors on the publication. An additional
co-author on this publication is Dr. Ali Almansoori who assisted in the direction of the
paper and provided feedback prior to journal submission.
iv
Abstract
This work develops a generalized modeling framework using several techniques for as-
sessing the feasibility of an eco-industrial network or ‘eco-park’ in order to demonstrate
the environmental and economic benefits of industrial facilities with cooperative goals to
conserve energy and materials. The work takes advantage of three distinct types of model-
ing techniques (linear programming, mixed-integer linear programming and mixed-integer
non-linear programming) to incorporate increasingly complex circumstances for designing
eco-industrial networks. The purpose of this research is to provide policy-makers and facil-
ity designers with an approach to optimize construction of facilities based upon economic
and environmental incentives. This framework allows for optimizing the material and en-
ergy efficiency of a network of facilities to reduce emissions, waste, and input of materials
and energy while maintaining production levels.
Major contributions from this thesis are to examine the potential for alternative-fuel
vehicles within the concept of a hydrogen economy and exploration of eco-industrial net-
works, utilizing the tools of life cycle analysis and system optimization. Life-cycle assess-
ment is utilized as a tool for decision-making throughout this thesis and is an invaluable
asset in making environmentally-conscious decisions. This type of assessment evaluates
the emissions of a product from virgin material extraction through to final disposition in
the aquatic, terrestrial or atmospheric domain. The use of life-cycle assessment techniques
shows clear impacts on society over the entire lifecycle of the products and processes con-
sidered herein. Development of a dual-objective function to account for economics and
environmental performance of industrial facilities is developed and utilized to aid in the
decision process for policy-makers and facility designers.
The concept of eco-industrial networks is further extended by including additional com-
v
ponents, such as transportation modes, within the model. To this end, preliminary work
examines the practical possibility of shifting automobile propulsion technologies to alter-
native fuels with emphasis on the criteria air contaminants considered herein of greenhouse
gases, volatile organic compounds, and oxides of sulphur and nitrogen. The scenarios pre-
sented are based on a model of the electricity system in the province of Ontario, Canada
and energy pathway analysis to assess the supportable market penetration of, and emis-
sions from, alternative vehicle technologies. The recommendation of this work is that a
transition to electric vehicles in the near-term followed by a transition to hydrogen fuel-
cell vehicles will yield the largest reduction in criteria air contaminants in both the urban
centre of Toronto, Ontario and in the province as a whole.
The consideration of transportation and transitional technologies feeds directly into
the concept of eco-industrial parks and the benefit to society of their implementation. The
reduction in transportation distance between relatable chemical manufacturers has been
hailed as a major benefit of implementing eco-industrial park topology. This work devel-
ops a generalized modeling framework for eco-industrial parks based on a dual objective
of societal and industrial requirements. The nodes considered in this work include: energy
generation via hydrocarbon gasification or reforming, carbon capture, carbon sequestra-
tion, pressure-swing adsorption in addition to the manufacture of ammonia and urea within
the context of refueling a fleet of 1000 hydrogen vehicles. Life-cycle assessment is applied
to form the societal benefits of operating facilities within an eco-industrial framework and
the long-term economics of the processes are considered to form the economic portion of
the objective. Modeling is carried out in three distinct types: linear programming, mixed-
integer linear programming and mixed-integer non-linear programming. Each of these types
represents a different modeling framework developed to assess various complexities in the
eco-industrial network and yet they share common goals, themes and analysis methods.
Using each of these approaches, a case study eco-industrial park is analyzed using the three
vi
types of modeling methodologies mentioned. The simpler LP model is unable to account
for some of the complexities inherent in an eco-park network and thus the results from this
model are subsequently viewed as an upper boundary on the benefits of eco-industrial in-
tegration for the case study mentioned. The subsequent efforts of mixed-integer linear and
non-linear programming serve to refine the model and provide more realistic investigation
of the benefits of such a network.
In order to achieve a reduction in emissions of harmful substances to the air, water
and land to meet national targets, analysis of the interactions between humans and the
environment must be explored to unlock new avenues of production and consumption to
reduce the impact that society is having on the environment. This work is completed
within the larger context of the potential hydrogen economy with the supposition that
such a scenario will be enabled by increasingly effective technology. The transition of our
current infrastructure to the hydrogen economy shows benefits to air quality from reduced
emissions of vehicles and also from a reduced industrial contribution.
vii
Acknowledgements
I would like to acknowledge my supervisors, Dr. Ali Elkamel and Dr. Michael Fowler,
for their patience and guidance in these studies; additionally, I would like to thank the
rest of my committee for taking the time to read my thesis and provide feedback on my
research work.
I would like to thank Bonnie De Baets, my partner, for her patience, understanding
and encouragement. Also, I would like to thank my friends and family for keeping me
grounded and involved in my community and life outside of the university.
I would also like to acknowledge NSERC and the Vanier scholarship program for pro-
viding financial support to pursue this endeavor.
viii
Contents
List of Tables xiv
List of Figures xvii
Nomenclature xviii
1 Introduction 1
2 Background 8
2.1 Eco-park Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 Examples of Eco-parks . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Academic Studies of Eco-parks . . . . . . . . . . . . . . . . . . . . 13
2.1.3 Industrial Examples of Eco-parks . . . . . . . . . . . . . . . . . . . 14
2.1.4 Geographical Differences in the Application of Eco-parks . . . . . . 15
2.2 Hydrogen Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Analysis Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Life Cycle Assessment . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.2 Typical Life Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
ix
2.4 Modeling Programs and Methodologies . . . . . . . . . . . . . . . . . . . . 24
2.4.1 Software Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5 Environmental Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.1 Climate Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.2 Air Pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.3 Ozone Depletion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.4 Acid Rain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.5.5 Resource Conservation . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.5.6 Habitat Destruction and Fragmentation . . . . . . . . . . . . . . . 30
2.5.7 Solid Waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.6 Financial Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3 Air quality and environmental impacts of alternative vehicle technologies
in Ontario, Canada 33
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.1.1 Health Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Modeling and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.1 Data Gathering and usage . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.3 Results of Supportable Penetration and Vehicle Growth . . . . . . . 44
3.2.4 Pollution Abatement Results . . . . . . . . . . . . . . . . . . . . . . 48
3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
x
4 Optimized production of hydrogen in an eco-park network accounting for
life-cycle emissions and profit 58
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.1.1 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.1.2 Eco-industrial Network Description . . . . . . . . . . . . . . . . . . 59
4.1.3 Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.1.4 Hydrogen Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.2 Network Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2.1 Chemical Processing Plants . . . . . . . . . . . . . . . . . . . . . . 61
4.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3.2 Definition of Environmental Metrics . . . . . . . . . . . . . . . . . . 72
4.3.3 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.4 Selection of Optimization Package . . . . . . . . . . . . . . . . . . . 76
4.3.5 Objective Function . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.4 Results of Reduced Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.4.1 Reduced Case Model . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.4.3 Hydrogen Optimization . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.4.4 Greenhouse Gas Emissions Compared to Criteria Air Contaminants 84
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
xi
5 Optimization of Material and Energy Exchange in an Eco-park Network
Considering Three Fuel Sources 88
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.1.1 Eco-park concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.1.2 Life-cycle Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.2 Manufacturing Facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Objective Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.4 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6 Generalized MINLP Modeling of Eco-Industrial Networks 120
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.2 Multi-Objective Optimization Model . . . . . . . . . . . . . . . . . . . . . 123
6.2.1 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.2.2 Problem Representation . . . . . . . . . . . . . . . . . . . . . . . . 127
6.2.3 Model Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
7 Summary and Conclusions 166
7.1 Summary of Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
7.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
xii
7.3 Summary Statement of Contributions . . . . . . . . . . . . . . . . . . . . . 172
7.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
7.4.1 Recommendations for Future Work . . . . . . . . . . . . . . . . . . 174
7.4.2 Recommendations for Implementation . . . . . . . . . . . . . . . . 175
References 176
Appendix 198
xiii
List of Tables
3.1 Calculated number of vehicles in Ontario and penetration for FCVs and
PHEVs for both transition scenarios . . . . . . . . . . . . . . . . . . . . . 46
3.2 Calculated number of each type of PHEVs and FCVs in Ontario and the
drivable distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.1 Table of input parameters for basic case in reduced model . . . . . . . . . 80
5.1 Legend of Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.2 Summary of inputs and outputs . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3 Comparison of life cycle emissions per kg of ammonia based on three fuels 118
5.4 Life-cycle emissions of ammonia based on three fuels at a reference level of
730 tonnes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.1 Capacities and reference costs of eco-park facilities . . . . . . . . . . . . . . 153
6.2 Summary of flowrates under pure economic consideration . . . . . . . . . . 155
6.3 Summary of flowrates under purely environmental consideration . . . . . . 156
xiv
List of Figures
2.1 Fundamentals of a product life cycle . . . . . . . . . . . . . . . . . . . . . 22
2.2 LCA Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1 Canadian greenhouse gas emissions by sector . . . . . . . . . . . . . . . . . 40
3.2 Overall electricity generation mix for Ontario . . . . . . . . . . . . . . . . 41
3.3 Base-load electricity generation mix for Ontario . . . . . . . . . . . . . . . 42
3.4 Transition scenarios for AFV penetration in Ontario . . . . . . . . . . . . . 43
3.5 GHG and CO2 reduction in Ontario . . . . . . . . . . . . . . . . . . . . . . 49
3.6 Normalized reduction in GHG emissions . . . . . . . . . . . . . . . . . . . 50
3.7 VOC emission reductions in overall and urban settings for AFVs . . . . . . 51
3.8 NOx emission reductions in overall and urban settings for AFVs . . . . . . 52
3.9 PM10 emission reductions in overall and urban settings for AFVs . . . . . 53
3.10 PM2.5 emission reductions in overall and urban settings for both types of
AFVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.11 SOx emission reductions in overall and urban settings for AFVs . . . . . . 56
4.1 Proposed network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
xv
4.2 Reduced network configuration representative of the reduced case . . . . . 79
4.3 Reduced configuration representative of the reduced eco-park . . . . . . . . 81
4.4 Results of producing hydrogen for 1000 fuel cell vehicles . . . . . . . . . . 82
4.5 Normalized plant exports with hydrogen production for 1000 cars for various
values life-cycle emissions weighting . . . . . . . . . . . . . . . . . . . . . . 83
4.6 Analysis of the impact of varying the weighting of greenhouse gas emissions
and criteria air contaminants with life-cycle emissions weighting of 0.5 and
profit weighting of 0.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.1 Visualization of an EIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.2 EIN considered in this work . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.3 Life-cycle GHG contributions for producing Ammonia . . . . . . . . . . . . 98
5.4 Life-cycle SOx contributions for producing ammonia . . . . . . . . . . . . . 99
5.5 Normalized plant exports as a function of life-cycle emissions weighting . . 112
5.6 Normalized network plant export using coal as gasification feedstock . . . . 114
5.7 Normalized network plant export using natural gas as feedstock . . . . . . 115
5.8 Network product exports and profit for three different fuels . . . . . . . . . 116
5.9 Network exports of power, heat and urea for three different fuels . . . . . . 117
6.1 Eco-industrial network flow diagram . . . . . . . . . . . . . . . . . . . . . . 125
6.2 Gasification process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.3 Combined heat and power process . . . . . . . . . . . . . . . . . . . . . . . 132
6.4 Carbon capture process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.5 Pressure swing adsorption process . . . . . . . . . . . . . . . . . . . . . . . 137
xvi
6.6 Ammonia process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.7 Urea process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.8 Optimal network schematic under purely economic consideration . . . . . . 154
6.9 Optimal network schematic under purely environmental consideration . . . 156
6.10 Results of the objective value for eco-park optimization varying objective
function weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.11 Results of variation in the objective value weights in two dimensions related
to weighting for life-cycle emissions . . . . . . . . . . . . . . . . . . . . . . 158
6.12 Normalized plant capacities under varying weighting of life-cycle emissions.
G is descriptive of gasification, CC of carbon capture, CHP for combined
heat and power, PSA for pressure-swing adsorption, SQ for sequestration,
AM for ammonia manufacture and U for urea production . . . . . . . . . . 159
6.13 Relative contributions of emission reduction objective and economic objec-
tive to the overall objective value considering weighting factors . . . . . . . 162
6.14 Relative strength of emission reduction and cost reduction factors on the
objective function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
xvii
Nomenclature associated with Chapter 3
Acronyms Description
AFV Alternative fuel vehicle
PHEV Plug-in hybrid electric vehicle
FCV Fuel cell vehicle
FCPHEV Fuel cell plug-in hybrid electric vehicle
HEV Hybrid electric vehicle
BEV Battery electric vehicle
OMA Ontario Medical Association
VOC Volatile organic compounds
GTA Greater Toronto area
NOx oxides of nitrogen
PM10 particulate matter of diameter less than 10 microns
PM2.5 particulate matter of diameter less than 2.5 microns
GHG Greenhouse gases
CO2 Carbon dioxide
SOx Oxides of sulphur
AMPL A Mathematical Programming Language
GREET Greenhouse gases, Regulated Emissions, and Energy use in Transportation
Variables Description Units /Type
γ time period
λi zonal base-load growth rate %
πi zonal peak-load growth rate %
λ annual base load growth rate of Ontario’s to-
tal zones
%
xviii
Variables Description Units /Type
bi base load value in Zone i MW
Phppiy the total installed HPP capacity in Zone i,
by Year y
MW
Psijy power component MW
Phiy required power of HPPs in Zone i by Year y MW
∆Phppiy newly installed HPP in Zone i and Year y MW
Thijy transferred hydrogen tonnes day−1
m number of trucks, integer
OCy operation costs of one compressed gas truck
in Year y
$CAD km−1
Pg zonal generation power in Zone i, Year y, and
during the time periodω1 orω2
MW
Pim imported power in Zone i, Year y, and during
the time periodω1 orω2
MW
Pex exported power in Zone i, Year y, and during
the time periodω1 orω2
MW
CCcab capital cost of cab trailers $
CCtube capital cost of tube trailers $
LTcab lifetime of cab trailers years
LTtube lifetime of tube trailers years
DR discount rate %
ntry maximum number of compressed gas trucks
in route between Zones i and j in Year y
integer
gij conductance of the line between buses i and
j
S
xix
Variables Description Units /Type
δ bus voltage angle rads
Plossi,j power loss in line i,j MW
αijy(l) slope of the lth block of voltage angle
deltaγijy(l) value of the lth block of voltage angle
bijy susceptance of the line (i,j) in Year y S
Pl total load in each zone MW
ε very small positive number none
thmijy auxiliary variable representing the trans-
ferred hydrogen
tonnes day−1
xx
Parameters Description Units
Pgiy
minimum zonal power generation MW
Pgiy maximum zonal power generation MW
Pimiy lower bound of imported power MW
Pimiy upper bound of imported power MW
Pexiy minimum exported power MW
Pexiy maximum exported power MW
Pdij maximum capacity of the transmission corri-
dor (i,j) in direct power flow
MW
Prij maximum capacity of the transmission corri-
dor (i,j) in reverse power flow
MW
Phppiy maximum size of HPP which is allowed to be
installed in Zone i by Year y
MW
dij approximate distance between Zones i and j km
xxi
Sets, Subsets and Indices Description
Z set of indices of zones or buses in the simplified
network
Z∗ is the set of indices of hydrogen transfer corridors
Υ1 set of indices of planning years starting in 2009
Υ set of indices of planning years starting from 2008
ω1 index for the time period corresponding to 8 week-
day hours (12 am-7 am);
ω2 index for the time period corresponding to 14
weekend hours (12 am-1 pm);
Ω set of indices of transmission lines
Ψ set comprised of ω1 and ω2
Binary Variables Description
Kmijy binary variable which takes the
value of 1 if m trucks are needed
for daily hydrogen transfer be-
tween Zones i and j in Year y
ωγijy(l) binary variable which takes the
value of 1 if the value of the lth an-
gle block for the line (i,j) is equal
to its maximum value ∆δy
xxii
Nomenclature associated with Chapter 4
Acronym Description
EIN Eco-industrial network
LCA Life-cycle assessment
LP Linear programming
GHG Greenhouse gases
CO2 Carbon dioxide
GAMS General algebraic Modeling System
G Symbol for the gasification unit
PSA Symbol for the pressure-swing adsorption unit
CHP Symbol for the combined heat and power unit
CC Symbol for the carbon capture unit
AM Symbol for the ammonia manufacturing unit
U Symbol for the urea manufacturing unit
ME Symbol for the methanol manufacturing unit
GH Symbol for greenhouses
SQ Symbol for the sequestration unit
EC Symbol for the energy crops
ET Symbol for the ethanol unit
Variable Description Units /Type
S Total sulphur output tonnes
SGisulphur output from each gasifier kmol
A Total ash output tonnes
AGiAsh output from each gasifier kmol
xxiii
Variable Description Units /Type
ySGijFraction of sulphur present in fuel j used
in gasifier i
Decimal
yAGijFraction of ash present in fuel j used in
gasifier i
Decimal
Fij Moles of fuel j used in gasifier i kmol
CO Total moles of carbon monoxide produced
from gasification
kmol
H2 Total moles of hydrogen produced from
gasification
kmol
N2 Total moles of nitrogen produced from
gasification
kmol
CO2 Total moles of carbon dioxide produced
from gasification
kmol
xij binary variable expressing whether fuel j
is being supplied to gasifier i
Binary
CO2pout CO2 output from plant p kmol
CO2pin CO2 input to plant p kmol
N2pout N2 output from plant p kmol
N2pin N2 input to plant p kmol
H2pout H2 output from plant p kmol
H2pin H2 input to plant p kmol
COpout CO output from plant p kmol
COpin CO input to plant p kmol
NH3pin Ammonia input to plant p kmol
NH3pout Ammonia output from plant p kmol
xxiv
Variable Description Units /Type
Mepin Methanol input to plant p kmol
Mepout Methanol output from plant p kmol
Upin Urea input to plant p kmol
Upout Urea output from plant p kmol
Epin electricity input to plant p kWh
Epout electricity input to plant p kWh
Hpin heat input to plant p MJ
Hpout Heat output from plant p MJ
γWGS Extent of water-gas shift reaction inside
the gasifier
Decimal
EG Energy converted in the CHP unit MJ
LHVCO Lower heating value of carbon monoxide MJ kmol−1
LHVH2 Lower heating value of hydrogen MJ kmol−1
HGX Heat generated for export MJ
EGX Electricity generated for export kWh
HeatSplit Split of energy conversion for heating Decimal
Z The objective value, value to society none
JLCE Contribution of life-cycle emission reduc-
tions to the objective function
none
Je Contribution of economics to the ojective
function
Dollars
ACCSp Amortized capital cost of plant p in an
independent operating scenario
Dollars
OCSp Operating cost of plant p in an indepen-
dent operating scenario
Dollars
xxv
Variable Description Units /Type
ACCIp Amortized capital cost of plant p in an
integrated operating scenario
Dollars
OCIp Operating cost of plant p in an integrated
operating scenario
Dollars
xxvi
Sets, Subsets and Indices Description
i plant index
j fuel index
p set of plants
e set of emissions
Ω set of fuels
Ωi set of fuels that can be utilized in gasifier i
Parameter Description units
Coste Environmental cost of emission e Decimal
FUi upper limit on supply of fuel to gasifier i tonnes
FUj Upper limit of supply of fuel j tonnes
n,m,k coefficients of a biomass product none
ηHG Efficiency of heat generation Decimal
ηEG Efficiency of electricity generation Decimal
WLCE Objective function weighting for life cycle emissions Decimal
We Objective function weighting for economics Decimal
xxvii
Nomenclature associated with Chapter 5
Binary Variables
xp,m =
1 if plant p of size m exists
0 Otherwise
yp,p2,k,v =
1 if material v is transported by method k between p and p2
0 Otherwise
xxviii
Acronym Description
EIN Eco-industrial network
LCA Life-cycle assessment
EIP Eco-industrial Park
GAMS General algebraic modeling system
MILP Mixed-integer linear program
DJSI Dow Jones sustainability index
WAR Waste reduction
ISO International Standards Organization
SOx Oxides of sulphur
CO2 Carbon dioxide
GHG Greenhouse gases
NPV Net present value
G Symbol for the gasification unit
PSA Symbol for the pressure-swing adsorption unit
CHP Symbol for the combined heat and power unit
CC Symbol for the carbon capture unit
AM Symbol for the ammonia manufacturing unit
U Symbol for the urea manufacturing unit
ME Symbol for the methanol manufacturing unit
GH Symbol for greenhouses
SQ Symbol for the sequestration unit
SMR Steam-methane reforming
xxix
Model Framework Description
Z Objective value
x Design vector
c Vector of fixed parameters
g vector of inequality constraints
h vector of equality constraints
xLi the lower bound for xi
xUi the upper bound for xi
q number of objectives
x vector of decision variables
c vector of fixes parameters
m1 number of inequalities; and
m2 number of equalities
xxx
Model Setup Description
λi weighting factor for objective Zi
Fi scaling factors for Zi
Z aggregated objective value
Z1 = Zeconomic economic portion of the objective value
Z2 = Zemissions environmental portion of the objective
value
ne number of emissions considered
λe weighting factor for emission e
Fe scaling factors for emission e
Ze emission differential between the inte-
grated and stand-alone facilities
Ie emissions of e from an integrated facil-
ity
Se emissions of e from a standalone facility
rd discount rate
np number of manufacturing facilities
Sets Description
k set of the transportation technologies available
m set of plant sizes
v set of material vectors
e set of emissions
p set of facilities; and
p2 alias of p
xxxi
Variables Description Units
outv,p amount of material v output from
plant p
kg; mol
inv,p input of v to plant p mol
rxv,p consumption of v in p mol
genv,p generation of v in p mol
MUp,m upper flowrate limit of from plant
p of size m
m3; kg; mol
Rp return from sale of products from
plant p
$
ICC integrated plant capital cost $
SCC standalone plant capital cost $
IOC integrated plant operating cost $
SOC standalone plant capital cost $
t year
L plant lifetime p years
TransportationCostk Total transportation cost of type
k
$
BaseCostk Base cost of transportation mode
k
$
ThroughputCostk Throughput cost of transporta-
tion mode k
$m−3; $kg−1; $mol−1
xxxii
Nomenclature associated with Chapter6
Acronyms Description
AHP Analytic hierarchy process
EIN Eco-industrial network
GAMS General Algebraic Modeling System
GHG Greenhouse gases
H total number of fuels available
LCA Life-cycle assessment
LCE life-cycle emission
LP Linear Programming
MILP Mixed-Integer Linear Programming
MINLP Mixed-Integer Nonlinear Programming
SMR Steam methane reforming
Continuous
Variables
Description Unit /type
ACCi,s annual amortized capital cost of the plant i with
scheme s
$ yr−1
ADu ammonia plant node’s utility u demand J h−1; tonne h−1; kW
AIm ammonia node’s material m input mol h−1
AOm,j ammonia plant node’s material m output to the
jth node
mol h−1
APm ammonia plant node’s material m product mol h−1
ARu ammonia plant node’s utility u requirement per
unit of NH3 produced
Jmol−1; tonnemol−1;
kWhmol−1
ASFj ammonia plant node’s split factor to node j %
xxxiii
Continuous
Variables
Description Unit /type
CCDu carbon capture node’s utility u demand tonne h−1
CCIm carbon capture node’s material m input mol h−1
CCOm,j carbon capture node’s material m output to the
jth node
mol h−1
CCSj carbon capture node split factor of the CO2-
deficient gas sent to the node j
%
CCSFj carbon capture node’s stream splitting factor to
the sink node j
%
CEC combined heat and power node’s input gas en-
ergy content
J h−1
CERu combined heat and power node’s energy ratio
used to produce utility u
%
CF problem’s objective cost function fractional
CGu combined heat and power node’s utility u gen-
eration
J h−1;kW
CIm combined heat and power node’s material m in-
put
mol h−1
COm,j combined heat and power node’s material m
(stack gas) sent to the node j
mol h−1
CPm combined heat and power node’s product m
generation
mol h−1
EPCi,s,e emission e related to the construction of plant i
with scheme s
tonnes emission e
xxxiv
Continuous
Variables
Description Unit /type
EPOi,s,e emission e from the operation of plant i with
scheme s
tonnes emission e
FM mass flowrate of the fuel entering the gasifier kg h−1
GDu gasifier’s utility u demands J h−1; tonne h−1; kW
GOm,j gasifier’s material m output to the production
node j
mol h−1
GPm gasifier’s material m production mol h−1
GSFj gasifier’s split factor to the sink node j %
GWm gasifier’s waste material m generation sent to
waste /water treatment facilities
mol h−1
MAMm,j market node’s material m available to be sold
to node j
mol h−1
MATERm,i,j material m going from node i to node j mol h−1; m3 h−1
MEAu,j market node’s utility u exported to node j Jh−1; tonne h−1
MEIu market node’s utility u input (i.e., utility sold
to the market)
J h−1; kW
MEOu,j market node’s utility u sold to node j J h−1; tonne h−1; kW
MERu market node’s utility u ratio input (i.e., utility
ratio that can be sold to the market)
%
MIm market node’s material m input mol h−1
MOm,j market node’s material m outputs to the jth
node
mol h−1
NEC eco-industrial network’s annual lifecycle emis-
sions comparison ratio
dimensionless
xxxv
Continuous
Variables
Description Unit /type
NPC eco-industrial network’s annual production cost
comparison ratio
dimensionless
NTCk networks transport cost for mode k $yr−1
OCi,s operating cost of plant i following scheme s $ yr−1
PCl,i,s plant capital cost $
PDu pressure swing adsorption nodes utility u de-
mand
J h−1; tonne h−1; kW
PIm pressure swing adsorption nodes material m in-
put
mol h−1
PlantCost2 plant scaled cost $
POm,j pressure swing adsorption nodes material m
output to the jth node
mol h−1
POCi,s plant’s operating capacity mol h−1
PPCl,i plant’s production capacity mol h−1; m3 h−1
PSj pressure swing adsorption node’s split factor of
the H2-deficient syngas to the node j
%
PSFj pressure swing adsorption node’s H2 stream
splitting factor to the jth node
%
SDu sequestration node’s utility u demand tonne h−1; kW
SIm sequestration node’s material m input from the
carbon capture node
mol h−1
UDu urea plant node’s utility u demand J h−1; tonne h−1; kW
UIm urea plant node’s material m input mol h−1
xxxvi
Continuous
Variables
Description Unit /type
UOm,j urea plant node’s output material m into node
j
mol h−1
UPm urea plant node’s product m mol h−1
URm,j urea plant node’s material m output to the jth
node
mol h−1
WDu waste/water treatment node’s utility u demand J h−1; tonne h−1; kW
WIm waste/water treatment node’s material m input mol h−1
Binary Variables
fh =
1 if fuel h is selected
0 Otherwise
xi,l =
1 if plant i of size l is selected
0 Otherwise
yi,j,k =
1 if transportation method k is used between node i and node j
0 Otherwise
Integer Variables Description
TUNk,i,j number of transport units k used to transfer materials from i to j
xxxvii
Sets and Subsets Description
d set of pipeline diameters
e set of environmental emissions considered in the LCE analysis
h set of feedstock fuels available in the eco-industrial network
i set of source nodes
j set of sink nodes
k set of transport modes available to connect the nodes
m set of materials (i.e., products and by-products of the EIN)
u set of utilities associated to the EIN
Set elements Description
ap ammonia production node
B biomass fuel
C coal fuel
cc carbon capture node
chp combined heat and power node
g gasification process node
GHG Greenhouse Gases emissions expressed in CO2 eq.
mk market place node
M1 carbon monoxide (CO)
M2 carbon dioxide (CO2)
M3 hydrogen (H2)
M4 nitrogen (N2)
M5 ammonia (NH3)
M6 urea CO(NH2)2
M7 water (H2O)
xxxviii
Set elements Description
M8 methane (CH4)
M9 ash
M10 sulfur (S)
M11 other contaminants
NG natural gas fuel
NOx oxides of nitrogen emissions
Pd transport pipeline
psa pressure swing adsorption node
Rd road transport
Rl rail transport
SOx sulfur oxides emissions
sq sequestration node
SW solid waste materials (i.e., landfill materials)
up urea production node
U1 heat utility
U2 process water utility
U3 electricity utility
wt waste /water treatment node
Model Parame-
ters
Description Units /Type
ACF ammonia plant node’s reaction conversion %
ACFi annual amortized capital factor associated
to the plant i
%
xxxix
Model Parame-
ters
Description Units /Type
ASFm ammonia plant node’s stoichiometric rela-
tionship for the production of material m
decimal
CCRu carbon capture node’s utility requirement
per CO2 product
tonnemol−1
CEFu combined heat and power node efficiency
for generating utility u
%
di,j distance between nodes i and j km
EFPCl,i,s,e emission e from the construction of a plant
of size l
tonnes emissione
EFPOi,s,e emission factor related to the operation of
the plants
(tonnes emission e)(h)
(yr)−1(mol)−1
FCh,m material m composition out of the gasifier
per fuel type h
mol%; mass%
FCk fixed cost associated with transport
method k
$ yr−1; $(km)−1(yr)−1
FCUk fixed cost per transportation unit type k $ yr−1
GRu gasifier’s utility requirement per unit of
product
J mol−1; tonnemol−1;
kWhmol−1
LHVm lower heating value of the gaseous compo-
nents m
J mol−1
MWh molecular weight of the feedstock fuel type
h
g mol−1
OCFi,s operating cost factor associated to the
plant i following scheme s
%
xl
Model Parame-
ters
Description Units /Type
PlantCapacity1 reference plant installed capacity mol h−1
PlantCapacity2 scaled plant installed capacity mol h−1
PlantCost1 reference plant cost $
POLi plant’s operating life years
PRu pressure swing adsorption node’s energy
requirement per unit of H2 product
J mol−1; tonnemol−1;
kWhmol−1
SRu sequestration node’s utility u requirement
per unit of CO2 input
kWhmol−1
TCk transportation method k’s capacity factor mol h−1; m3h−1
UCF urea plant node’s reaction conversion %
URu urea plant node’s utility u requirement per
product
J mol−1; tonnemol−1;
kWhmol−1
USFm urea plant node’s stoichiometry for the
production/consumption of material m
unitless
V Ck variable cost of the transportation method
k
($)(h) (km)−1 (yr)−1 (mol)−1;
($)(h) (yr)−1 (mol)−1
WE weight assigned to the lifecycle emissions
of the network
%
WP weight assigned to the production cost of
the network
%
WRu waste /water treatment node’s utility re-
quirement per tonne process water
J tonne−1;
kWh tonne−1
xli
Chapter 1
Introduction
This work focuses on demonstrating the benefits of eco-industrial integration with respect
to environmental and economic benefits while providing a holistic production of heat, elec-
tricity, industrial products and transportation fuel. Sections of this work contribute to the
optimization of a network of chemical facilities considering economics and emissions from
virgin material extraction to the network boundary, while another portion is focused upon
the impacts of alternative transportation fuels from a complete life-cycle perspective. The
scope of this work is limited to economic assessment and four environmental impact cate-
gories, namely: climate change potential, acidification, urban air quality and solid waste.
This work is intended to impact the areas of facility design/construction, environmental
assessments of industrial operation and to a large degree, those who create and influence
policy within the industrial sector. Balance between economic viability and environmental
impact for the operation of industrial facilities are balanced throughout this work; there-
fore, the most relevant recipients of this work are those who would influence the policy
decisions for reducing environmental impact from industry without discouraging invest-
ment in construction of such facilities. The balance between economic and environmental
components as part of a bi-objective optimization in eco-industrial networks (EINs) is a
1
novel method and unique contribution to this field.
This thesis will contribute to re-invigorating the concepts of eco-industrial integration
by empirical analysis of potential economic and environmental benefits. Implementation of
the proposed modeling frameworks across a network of chemical facilities including syngas
production, combined heat and power, carbon capture, carbon sequestration, ammonia
manufacture, pressure-swing adsorption and urea production is completed to show the in-
tegration framework and possibilities for economic and environmental benefit from this
set of facilities. The selection of facilities was completed with the mindset that integrated
systems should have commonalities in the form of energy and material vectors utilized,
endothermic/exothermic coupling to maximize the potential integration of material and
energy transfers throughout the network.
Chapter 2 describes the methodologies utilized throughout this thesis with emphasis
placed on life-cycle assessment (LCA) and eco-industrial network (EIN) concepts. The
scope of this work is to provide the reader with the concepts necessary to comprehend the
work contained in Chapters 3 - 6 and to provide context for this research. Some of the
topics contained in this chapter include modeling techniques and programs used through-
out this work, the hydrogen economy context for this research and the environmental and
economic metrics discussed throughout this thesis.
Chapter 3 explores the positive contributions to reducing urban air pollution in the
greater context of the hydrogen economy and alternative transportation options. This
work examines the supportable market penetration of alternative-fuel vehicle types within
the confines of the electricity grid in Ontario, Canada. ‘A Mathematical Programming
Language’ (AMPL) software is utilized in the preliminary model to explore the support-
2
able penetration of these vehicles technologies given the constraints of the electricity grid.
The second part of the analysis is connecting these supportable penetrations to a reduction
in emissions in both the urban setting of Toronto, Ontario and in the overall context of
Ontario, Canada. GREET 1.8b software, developed by Argonne National Laboratories in
the USA is utilized in order to assess the emission reductions from each of the alternative
vehicle technologies. The overarching goal of this chapter is to analyze the potential im-
pacts of adopting these alternative fuel technologies related to the air quality in the urban
centres in the province of Ontario. Emissions from constructing hydrogen fuel distribution
infrastructure are not included in this work as they are external to the fuel production
and its use in the vehicles. Additionally, this chapter is based on current operation of
the electrical grid in Ontario including planned improvements for generation, the emission
calculations for electricity are based on the current mix of generation used in Ontario. This
work includes ‘cradle-to-grave’ analysis of alternative fuels used in vehicles pertaining to
the areas of climate change and urban air pollution. This chapter establishes a baseline re-
quirement of hydrogen generation for 1000 fuel cell vehicles persisting in the eco-industrial
integration scenarios in later chapters.
Chapter 4 continues the emphasis on emission reductions from Chapter 3 but applies
the combined analysis of economics and emissions to an optimization framework of an eco-
industrial park. This chapter attempts to connect the benefits from reduced transportation
emissions to the chemical manufacturing industry in order to reduce the emissions from
a group of chemical production facilities while providing the fueling needs for a limited
number of fuel cell vehicles, the benefits of which were shown in Chapter 3 in terms of
reduced airborne emissions. The modeling framework for this analysis is presented as a
linear programming (LP) model for a group of facilities with the end goal of improving
profitability while reducing life-cycle emissions for the end products. The case study of
plants explored in this chapter yield hydrogen, ammonia and urea with reduced emissions
3
when compared to conventional production. The network of facilities to be considered in
this work includes syngas generation (by coal or biomass gasification or by steam-methane
reforming of natural gas), carbon capture, carbon sequestration, combined heat and power
production, pressure-swing adsorption as well as the manufacture of ammonia and urea.
This analysis is also complete in the context of a potential hydrogen economy where hy-
drogen becomes a valuable product to support the transportation network. Two of the
primary benefits connecting this work with the work in Chapter 3 are the reduction in
emissions by operating hydrogen fuel-cell vehicles and the reduction in transportation dis-
tance for the feedstock materials used by the production facilities. The CPLEX solver
within GAMS is utilized for solving this LP and utilizes a simplex-based algorithm to de-
termine a global optimum for this model. This work is a ‘cradle-to-gate’ assessment for
the facilities mentioned and as such, it is assumed that any usage beyond the network
boundary will have an emissions profile identical to current usage, thus offering no benefit
beyond this boundary. Benefits from alternative usages beyond the network boundary are
described in Chapter 3 for hydrogen and electricity utilized as transportation fuels and
exhibit the benefits of this application.
The work in Chapter 5 expands the modeling framework from the LP model discussed
in Chapter 4 by including integer and binary constraints such as decisions for existence
of facilities and connections. The mixed-integer linear programming (MILP) model that
results from this expansion includes the ability to assess multiple hydrocarbon fuel sources
(specifically biomass, coal and natural gas) for manufacturing the chemical exports from
the eco-industrial park. The expansion of the modeling efforts to the MILP domain was
considered necessary as the LP model was limited in its practicality for large industrial
systems. This work contributes to the goal of assessing the economic and environmental
reductions which can be experienced by operating facilities in an interconnected network
of chemical plants and furthers the quantification of these benefits. By analyzing decisions
4
with added modeling flexibility of binary and integer constraints, the model provides addi-
tional realistic constraints on the construction and operation of an eco-industrial network.
The CPLEX solver within GAMS is again implemented to provide a global optimum for
this model, though the CPLEX algorithm for solving MILP models is based on branch-and-
bound techniques to accommodate the integer and binary portions of the model combined
with the simplex method for solving the constituent linear problems.
Chapter 6 takes the analysis from Chapters 4 and 5 to the next level of modeling com-
plexity by incorporating non-linear constraints into the MILP model. This mixed-integer
non-linear programming (MINLP) model is a much more complex optimization problem
and requires alternative solver algorithms and computational resources when compared to
an MILP or LP model. This step toward MINLP was necessary as the additional complex-
ity required from the constraints could not be forced into a linear form as it was previously
with the LP and MILP models. The goals for this framework were to create the most com-
prehensive optimization model with the same case study facilities explored in Chapter 5
but utilizing a more sophisticated model in an attempt to provide additional accuracy
and realism to the model. The models presented in Chapters 4 and 5 were completely
redeveloped with the goal of a relativistic objective function and increased realism for
constraints placed on the optimization. The framework developed for the MINLP model
serves to investigate the benefits of operating chemical facilities under stringent economic
and financial constraints. The model in this chapter also applies the most conservative
estimates on the benefits of the eco-industrial park case study and further investigates the
relationship between economics and reduction of environmental pollutants. The BARON
solver in GAMS is used to solve the MINLP model in this chapter and provides a global
optimum under general bounding assumptions for the model variables. The BARON algo-
rithm implements a deterministic solution approach and draws upon additional MILP and
NLP solvers making use of a branch-and-bound approach to return a global optimum.
5
Chapters 4, 5 and 6 assess environmental benefits from reducing ‘cradle-to-gate’ emis-
sions with the assumption that emissions beyond the network boundary will be equivalent
to those from conventional production. Chapter 3 explores the impacts of the entire life-
cycle of alternative fuels in vehicle applications on both urban and gross air emissions. The
environmental metrics used in this thesis are reflective of major environmental concerns
such as climate change, acidification and urban air pollution. The form of the objective
function utilized in Chapters 4, 5 and 6 can be used by policy-makers to influence decisions
in facility siting, design, construction as well as to work with corporations to achieve a re-
sult that reduces environmental impact from operations with marginal economic impact.
This allows for a collaborative relationship between industry and regulators to assure a
net-benefits solution for all participants.
Chapter 7 draws upon the work completed in Chapters 3 - 6 and summarizes the conclu-
sions of this work, the contributions and significance of the research and recommendations
for implementation of the results and methods explored herein.
The progression through each of the modeling types exhibits the benefits and limita-
tions of each methodology. These methods are assessed for their usefulness in modeling
EINs and express how each methodolgy can provide specific, explicit results. The objective
function is modified through each model developed for the eco-park under consideration
and expresses the balanced approach between economics and emissions explored through-
out this work. The specific examples shown through the model formulation are for the
proposed EIN, yet the generalized modeling framework developed by this work can be
applied across a broad range of integration scenarios for many types of facilities and situ-
ations. The introduction of additional modeling complexity to the formulation makes the
6
MINLP a complex, yet valuable tool in assessing potential benefits from operating many
facilities within an eco-park network compared to stand-alone facilities.
The methodologies developed herein are powerful implementations of facility optimiza-
tion that could be used by policy-makers to suggest approaches toward industrial develop-
ment and setting realistic goals for reduction of virgin material usage and wasted energy.
The same framework can be applied by industial entities to achieve such policy goals
such as a reduction in environmental pollutants while improving on the business goals of
profitability and corporate sustainability.
7
Chapter 2
Background
This chapter provides background on the concepts and methodologies used throughout
this research. This chapter contains background information on eco-industrial networks,
analysis tools such as life-cycle assessment, the hydrogen economy, modeling methods and
programs used in this work and environmental and economic metrics.
2.1 Eco-park Concepts
Eco-park concepts are several decades old and rely on collaboration from progressive fa-
cility managers in order to implement a symbiotic strategy for responsible and sustainable
chemical processing. Examples of these concepts are available in several European coun-
tries and are also found sparingly throughout North America and other parts of the world,
although the major drivers for these collaborative efforts are typically economic.
In this work, a number of industial chemical manufacturing and energy conversion
facilities are envisioned as working cooperatively to share outputs, emissions and wastes
8
within a network in order to improve profitability and overall environmental impact. Such
a network of the facilities can be termed an ‘eco-park’, eco-industrial park (EIP) or eco-
industrial network (EIN). Generally, materials and energy are exchanged between eco-park
partners such as:
• electricity;
• heat;
• fuel gases such as natural gas or hydrogen;
• base organic chemicals such as methane, methanol, DME, ethylene, benzene, organic
chlorides, solvents, etc.;
• base inorganic chemicals such as sulphuric acid, inorganic chlorides, hydrogen, hy-
droxides, etc.; and finally,
• water and waste.
An eco-park will ultimately seek to improve profitability of the network as a whole.
Key to the success of such a network is that the network profitability is apportioned ap-
propriately among the various facilities such that each facility recognizes the benefit of
participation in the EIN. Regulators, as the agent for society, can encourage the formation
of an ecopark through regulation, emission costs, solid waste disposal fees, etc. Regulators
can use incentives or fees (i.e., fees, taxes, fines, levies, etc.) in order to achieve the desir-
able outcomes. Especially in the case in which fees such as environmental levies or taxes
are to be avoided, eco-park principles can be applied to reduce the number or severity of
these measures.
One major focus of an eco-park should be the recovery of waste heat and allocation of
heating. Practically every chemical processing plant and almost all buildings in Canada
9
require heating, and often produce or consume significant amounts of process heat. Each
chemical plant has its own reservoir to discharge waste heat, which results in massive en-
ergy waste in each facility. If locations were close enough, heat and steam supplies could
be shared between facilities and would result in massive operational savings by optimizing
the design of the system [1]. The operational savings of this plan are not the sole factor
on which an eco-industrial system is based. In this analysis, the eco-park is assumed to
be in the developmental and/or planning stage, heat/steam supply and distribution can
be incorporated directly into the design of each respective plant. This is not to suggest
that this type of arrangement could not be constructed as a retrofit, merely that it would
reduce capital costs if done during the initial build. In fact, there have been several in-
stances of retrofit heat/steam handling cooperation, most recently in parts of Alberta near
Edmonton which is the oil-refining hub of Western Canada, including the oil sands refining
processes. Heat distribution is also one of the key factors in the Kalundborg network (dis-
cussed in section 2.1.1), where a single facility provided heating and power for a number
of neighbouring sites. Waste treatment and remediation is another obvious benefit of de-
veloping an eco-park. A large majority of chemical plants have waste storage facilities and
remediation plans specific to the particular process. The cost of construction and remedi-
ation typically follows a ‘base cost plus’ pricing scheme having a fixed cost for equipment
rental and design, with a comparatively smaller incremental sum depending on the size of
the site to be constructed/remedied. Thus, pooling the industrial waste between several
parties would significantly reduce the economic burden on each individual facility. Also,
due to strict environmental regulations being enacted, remediation of these sites must be
evaluated prior to any preparation of the site for construction and must be budgeted for
in advance such that funds are available to recover the natural ecology of the land. Again,
distributing this burden among several parties would greatly reduce the financial burden
on any particular party.
10
Reduction of waste is a major focal point for many eco-park plans and exchanging
co-products between facilities is one method, and the most intuitive, for accomplishing an
overall reduction in eco-park waste [2, 3]. Reduction of industrial waste was the primary
premise behind the Kalundborg cooperative [4, 5, 6]. The oil refinery acted as the central
hub for the system, and its waste streams were used by neighbouring facilities that also
took part in this relationship. Waste sulphur, normally a major environmental constraint
in a refinery, was used to make gypsum in a nearby facility where it was processed into
drywall. Waste gas and water from the refinery were used in the afore-mentioned power
plant as fuel gas and cooling water. Excess steam from the refinery and power station were
used for heating in all of the nearby buildings including several greenhouse installations.
The waste gas from the refinery was not sufficient to fuel the entire power station, yet
the coal fly ash from the adjacent power station was captured and processed into cement
in a nearby facility. Each addition to the network made it more economical and more
environmentally attractive for the citizens of Kalundborg. Significant cost savings were
experienced by each partner in the cooperative; thus, it was feasible to implement without
losing profits. Developing initiatives without incentive for industrial partners have a very
small chance of being supported by large industries, as most will not sacrifice profit to
improve the environment without clear benefit to the shareholders [7].
2.1.1 Examples of Eco-parks
Kalundborg, Denmark is the most recognized eco-park as it was one of the first applications
of eco-park concepts that was developed and studied [4, 5, 6]. The network began almost
accidentally as the business operators suggested that it might simply make sense to start
conducting business in a different manner. This example of eco-park concepts invigorated
the search for other potentially symbiotic sites and has served to show the potential bene-
11
fits of industrial collaboration.
Kalundborg has had incredible success with its industrial operations and has increased
profits dramatically among its industrial stakeholders. As wasted resources are inherently
uneconomical, the Kalundborg network manages to save 1 million cubic meters of ground
water each year, reduce annual oil consumption by 20 000 tonnes and has created mar-
kets and alternate uses for countless other normally wasted materials [5]. These massive
reductions in waste, in addition to being incredibly beneficial for the environment, have
also significantly boosted the profits of each organization involved. Economic and envi-
ronmental benefits must be exhibited in order for industry to have renewed interest in the
concepts of EINs. This information is available [8, 9, 10]yet the lack of published examples
throughout the industrialized regions of North America are proof that the information is
not known or there is some barrier to implementing these techniques. Jacobsen et al. [11]
presents concepts for mobilizing industrial symbiosis and sustainability for a variety of
industrial settings.
North American examples are considerably fewer than those in Europe as eco-park con-
cepts have yet to root themselves amongst business owners and leaders. As mentioned in
Section 2.1.3, most of the work in the area of North American eco-parks has been limited
to the petrochemical industry as it is one of the most prolific and centralized industries
available. As many of the co-products from organic chemical manufacturing can be reused
or reprocessed in a refinery, applying eco-park concepts to these processes tends to be very
advantageous for the industries involved. Although many of the proposed networks have
yet to become fully functional, the opportunities are present and academic studies have
shown that there may be great benefit in their operation [12, 9].
12
An additional example appears in the Brownsville/Matamoros park which is a joint
project between industry in Mexico and the United States with the major products of
cardboard, plastics, automotive products, oil and solvents. Examples elsewhere in the
world are limited due to challenges presented by developing nations [7]. A proposal of a
network in Rio de Janeiro, Brazil exhibits the fact that this type of network is possible in
developing nations and that these areas may benefit the most from such an integration [13].
2.1.2 Academic Studies of Eco-parks
The concept of eco-parks has been studied in the academic setting as a practical way for
industry to mimic the symbiotic effects of the natural world. Studies in this field have often
concentrated on the petrochemical industry as there are many possibilities for exchanging
co-products between refining facilities and manufacturers of plastics, paints, solvents, pro-
pellants and others. The theory behind eco-parks is an overall conservation of materials and
energy which has proven to be of interest in many fields of study in the academic setting.
While some studies have quantified economic benefits from these industrial integrations,
many produce qualitative analysis of the possibilities without empirical assessment. The
analysis discussed in this work serves to provide a quantitative approach to calculate the
environmental and economic benefits from integrating industrial facilities to form an eco-
park. This method can provide empirical evidence to support these cooperative initiatives
from both a financial and environmental position. Previous academic pursuits in this area
have failed to encompass all of these aspects using one technique [14, 15, 16] but do provide
insight into the potential ecological and economic benefits of integration which are useful
when attempting to create metrics for optimization as described in section 2.5. The aca-
demic realm of planning and evaluating the performance of eco-parks has a commonality
in the fact that it is a diverse field and requires a multidisciplinary approach in order to
13
properly assess the arrangement and operation of networked facilities [17].
Cote et al. [18] discusses, on a qualitative level, how eco-parks may be designed and
give several examples of projects being considered but no mention of any quantitative ev-
idence or optimal design to maximize profits or emission reductions. Cote et al. [19] also
discusses environmental assessment of small enterprises in Canada, choosing to develop a
new system of metrics but with limited results presented of the analysis.
Monteiro et al. [20] explores two routes for producing dimethyl carbonate (DMC) via
ethylene oxide or urea methanolysis but acknowledges the limits of attempting to optimize
a system within the constraints of HYSYS. Furthermore, the analysis does not account for
exchanges between this process and others, as an eco-park optimization would. Similar
ideas can be found elsewhere [21] but lack the supporting optimization complexity to fully
support decision-making.
2.1.3 Industrial Examples of Eco-parks
Industrial applications of eco-park concepts can be seen to some extent in Eurasia, Ocea-
nia and, to a lesser degree, in North America [7]. Investigations into applying eco-park
concepts in developing countries has not been ignored and may prove to be one of the
most cost-effective methods for industrializing these nations [22]. The European eco-park
concepts depend heavily on co-location and process similarities while most published eco-
parks in North America are purely petrochemical. An excellent example of this is a thesis
from Louisiana State University on the integration of facilities located in the Louisiana
petrochemical corridor [23]. Similar concepts can be seen in areas of oil production and
importation. Non-petrochemical eco-parks have been developed to a lesser extent world-
wide but have the potential to conserve vast amounts of fossil fuels, raw materials and
14
energy.
Networks of companies were formed in the late 1990s and early 2000s due to the interest
in industrial cooperation for the benefits that may be realized from such symbiotic rela-
tionships. One example of this is the Canadian Eco-Industrial Network (CEIN) which was
developed in 2000 but has not been active since 2005. The legal implications of commod-
ity trading between co-located facilities is a barrier to implementation as failure to deliver
products in a timely manner and in the specified quantities can lead to operational upsets
within any chemical facility. Although applications of these principles can be seen, it is
generally simpler for facilities to run as individual cells instead of as a network of processes
for the ease of operations, legality and communications. European facilities, however, have
shown that integration can lead to financial benefits which have drawn more interest in the
industrial community. EIN development has drawn much more interest from the academic
realm than it has in the industrial setting. In North America, one of the most major
barriers to implementation is the shipping of products between facilities. Gibbs et al. [24]
presents the planning of eco-parks within North America and observes some critical points
about this practice. Evaluating eco-parks can also have unique issues between countries,
eco-parks and facilities [12, 25].
2.1.4 Geographical Differences in the Application of Eco-parks
The North American climate for eco-parks is somewhat more limited by geography than are
eco-parks in Europe due to the vast distances between producing facilities. As Canada and
the United States are both very large countries with populations spread across thousands
of kilometres, integration of chemical processing facilities can be financially challenging on
a simple basis of shipping costs and constraints. The location of facilities is a related topic,
typically with a seperate optimization to determine the location of various plants with
15
respect to other facilities and proximity to natural resources [26]. North American rail
networks are also less developed than their European counterparts, leading to increased
dependence on roadway transportation with other considerations to pipelining and trans-
portation by ocean. Marine transportation is limited to port cities while pipelining is
intrusive to the land and is limited to fluid applications. The effects of each type of ship-
ping can be assessed in terms of cost and its impact on the air, water and land. This
transportation cost within an eco-park, both economic and environmental, must also be
weighed with the associated costs of production.
2.2 Hydrogen Economy
The ‘hydrogen economy’ is considered to be the next generation in energy infrastructure, in
which hydrogen is used as an energy vector for many applications which currently employ
fossil fuels. Hydrogen is easily stored and can be combusted or utilized in a fuel cell to
provide electricity. Hydrogen also has the benefit of flexible production using a variety
of fuel sources such as hydrocarbon reforming or water electrolysis. The only product of
hydrogen combustion is water as is also the case for using hydrogen in a fuel cell; thus, it
is considered to be a cleaner fuel than many alternative energy vectors. For motive power,
hydrogen can be employed in a combustion engine or fuel cell within a vehicle to provide
power to the engine with the exhaust composed of simple water vapour. Many major
vehicle manufactures will commence marketing a fuel cell vehicle in 2015. In electricity
systems that desire peak-shaving, load-leveling or peak-shifting, hydrogen can be produced
in off-peak times and stored to be utilized during high-demand periods to produce elec-
tricity. This method of energy storage is particularly useful in regions such as Ontario,
Canada where there is a high baseline production of electricity from carbon-free nuclear
plants, allowing for hydrogen production by electrolysis during low-demand periods. The
concept of ‘power to gas’ in which hydrogen is produced and then distributed and stored
16
within the existing natural gas infrastructure is also explored in the literature [27].The hy-
drogen economy also has several other benefits, one of the primary ones being to encourage
energy independence in countries that currently rely on imports of fossil fuel from unstable
economies to meet the demand for energy. Additional social and political reasoning for a
transition to the hydrogen economy are discussed in the literature [28, 29, 30], this work
is focused on operation within the current infrastructure but connects exceptionally well
to a scenario of the hydrogen economy.
2.3 Analysis Tools
2.3.1 Life Cycle Assessment
Key to the evaluation of the performance of industrial facilities is the concept of life cycle
assessment. Life-cycle assessment (LCA) is a methodology developed to account for the
impacts of a product, process or service over its entire lifetime from the initial extraction
of virgin materials to the final disposition into the air, water or land.
LCA is defined by the ISO 14040 standard in the following way: “LCA is a technique
for assessing the environmental aspects and potential impacts associated with producing a
product” [31]. As such, LCA can thus be seen as a tool that is to be used as a part of an
environmental management system (EMS) in order to improve the quality of the practices
within a company with respect to sustainable development and the environment. LCA can
also be referred to as a “Cradle-to-Grave” assessment as it incorporates the environmental
effects from the initial stages of extraction to its final disposition.
In this research, it is desired to apply LCA metrics and methods to optimize a system
17
of production processes in order to optimize the materials and energy usage. One typical
use for LCA is to compare alternative processes to determine which has the least life cycle
impact on the environment [32, 21]; therefore, this methodology is very much in agreement
with life cycle principles and applications.
2.3.1.1 History of LCA
The first recorded usage of an LCA-like methodology was by Harold Smith in 1963 for
the World Energy Conference of that year. Since that time, LCA has seen large shifts
in popularity due to changing economies and societal demands. One of these shifts was
experienced during the economic turmoil that followed the crash of the oil markets in the
1980s at which point public and private concerns were more focused on recovering from
the economic shock. When the “green shift”’ began in the early 1990s, LCA again be-
came favourable and started to be incorporated as a key tool and strategy in management
systems. Standardization of LCA began in the late eighties and early nineties [33] and
was mainly implemented to curb rampant misuse of the methodology leading to incor-
rect advertising statements and social views of companies. The first workshop on LCA
was held in Smugglers Notch, Vermont and was organized by the Society of Environmen-
tal Toxicology and Chemistry (SETAC). The concepts of the life-cycle inventory, impact
analysis and improvement analysis were founded during this workshop. A study on milk
packaging in Europe in 1990 showed the necessity of standardizing the LCA methodology
as the different methodologies led to very different results for the study [34]. The varying
studies showed exhorbitant variation in pollution and solid waste production per container
produced, causing LCA to be criticized as an ineffective method to account for waste.
The weaknesses of the varying studies were identified and specifically addressed in order to
achieve standardized methods which act as a base for past, current and future assessments.
18
2.3.1.2 Uses of LCA
A company may pursue life-cycle analysis and management within its operations or product
management for a number of reasons. The most common reason for this is that the company
is using an environmental management system and that LCA is simply a tool within this
management system [4]. As such, it provides measurable metrics in order to demonstrate
improvement toward objectives and targets; thus, it integrates very well with the process
and policies already in place. If a company is not already using an EMS, they may choose
to use LCA or a life-cycle mentality as a way of evaluating their environmental impacts
on a purely altruistic basis; typically, though, there is an end goal or benefit in mind
and that is to demonstrate environmental stewardship to society for the associated public
and employee relations benefits [35]. One such benefit of using LCA is that the company
may be able to use the results as part of a cost/benefit analysis, where the environmental
impacts are weighed as benefits against the cost of a certain project, product or process.
Using LCA also allows companies to advertize and interact with potential consumers by
showing their commitment to life-cycle thinking and to the environment. Finally, LCA
is a practical tool to measure improvement of materials and energy usage and such an
improvement contributes to the profitability of the operation as a whole. Reduced costs
can be realized in a number of areas such as:
• reduced material costs;
• reduced energy costs;
• reduced hazardous material management costs and hazardous waste disposal costs;
• reduced solid waste tipping fees;
• reduced waste and emission treatment costs;
• lower lost production time associated with fugitive releases and plant shutdowns;
19
• better regulator tracking;
• reduced fines and fees associated with environmental incidents; and,
• better relations with regulators and insurance providers.
2.3.1.3 LCA Process — Establishing a Baseline
Before implementing a change to a product or process and evaluating it from a life-cycle
perspective, it is generally advised to conduct a preliminary study in which the baseline for
the current process is established. This is an important step since the improvement may
cause the emissions from the plant under consideration to decrease but may increase the
overall impact on the environment due to upstream processing. One example of this would
be to consider a new catalyst which would increase the single-pass conversion from 80% to
90% but its production requires 50% more energy and emits large amounts of toxic agents
to the environment. Because LCA encompasses the impacts from all stages of manufacture,
adoption of the new catalyst can be assessed based on its full impact throughout the supply
chain.
2.3.1.4 Legislated Commitments
Currently, LCA is not required by specific legislation; however, emissions regulations, es-
pecially in the developed nations, are becoming increasingly stringent as the technology is
developed to curb these emissions and society demands that industry consider impacts on
the environment and potential impacts to the health of humans. Obeying policy is thus an
atypical reason for applying LCA methodology to date and thus LCA completed to date
would be by company mandate rather than regulator mandate.
20
2.3.2 Typical Life Cycle
The areas of interest for LCA are generally from the extraction of raw materials to the
production of the intended product and the associated product disposal into the environ-
ment via air, water or land. The process or product under consideration can generally be
visualized as undergoing the steps shown in Figure 2.1. This figure demonstrates that raw
materials and energy are inputs into the product, these materials are processed using energy
and finally are recycled or disposed of as waste. The outputs from each step are emissions
to air, land and water as well as the desired product and its associated co-products. Figure
2.1 illustrates that upon extracting the raw materials from the environment, these raw
materials are processed into a final product. Typically, this processing includes several
steps of manufacturing before the eventual product is created. This upstream manufac-
turing generally includes several steps of bulk processing which produce bulk feedstock
for many different applications. These bulk processors commonly produce co-products in
addition to the desired material as basic feedstock for other processing facilities; therefore,
allocation of emissions in these instances is extremely important and must be considered
appropriately.
Once the appropriate feedstock has been created, it is processed into the finished prod-
uct within the facility in question. Following its manufacture, the product is packaged
and then transported to the appropriate customer as required. The product is then used,
re-used and maintained until the user deems it time to retire the product and leads the
product to its final disposition. The final disposition of a product should not be treated
as being synonymous with land-filling, as there are many options for disposition including
recycling and incineration in addition to the option of land-filling. LCA is an integral part
of an environmental management system (EMS) as a tool that can be used to evaluate op-
tions and future projects in terms of environmental/sustainability metrics, can be included
as part of a cost/benefit analysis and can also be used to identify bottlenecks within a
21
Figure 2.1: Fundamentals of a product life cycle, adapted from ISO Standards [31]
system that currently exists.
2.3.2.1 Typical LCA
Figure 2.2 shows the steps to completing an LCA. The first stage is to determine the
scope of the project which includes identifying project goals, study specificity, methods
of data collection and timeline. The definition and scoping steps are typically reviewed
as the project progresses to ensure that the goals are being followed and that any devi-
ations must be noted and are still in accordance with achieving the goals of the assessment.
The second step of LCA, inventory, is generally the most time-consuming portion of
22
Figure 2.2: LCA Framework, adapted from [31]
an LCA and tends to also be the most difficult step in a life-cycle study. The participants
need to collect all of the appropriate data in this step after revisiting the methods of data
collection, boundaries, specificity and relevance.
Third, the data is analyzed under life cycle impact assessment procedures. The infor-
mation collected during the inventory is evaluated systematically and it is determined how
each impact category, such as climate change or ozone depletion, will be affected by the
product or process emissions. These impact categories are defined in the goal definition
and scoping and can reflect any of the concerns that are brought forward regarding poten-
tial impacts of a product or process. These impact categories are ranked according to their
importance in the study and thus the full effects of production can be assessed according
to these impacts.
The final block of the LCA framework is interpretation and monitoring. This portion of
23
LCA serves to provide checks for each step of the analysis in order to ensure the goals are
being met and that the analysis is being conducted according to the proper procedures.
The evaluation of process modifications is also carried out after the changes have been
made. The modifications are evaluated based on the expected performance and the actual
results achieved by collecting additional data to monitor the success of the change.
2.3.2.2 LCA in this Work
LCA is referred to as a major part of this work as the goal is to evaluate the viability
of eco-parks based on life-cycle principles [4]. Process improvements that simply allocate
product emissions to another processor cannot be considered as benefits for the process.
The cost, energy and emissions associated with removing an impurity, for example, must
be done at some point along the production pathway. If one facility in the process has
observed that the impurity does not affect their operation, they may choose to simply pass
the issue to the recipient of that chemical who must then remove the impurity as it can
damage a critical system. By ceasing the removal of the impurity, the former company may
show results that their product now uses less energy and is less harmful to the environment
because of the decreased emissions from their plant. In reality, the emissions have still been
produced and only the location of these emissions has changed. Evaluating the entire eco-
park on LCA principles exhibits the consequences for all of the network facilities while
avoiding the superficial appearance of reduced emissions where this is not the case.
2.4 Modeling Programs and Methodologies
The modeling in this work is completed using both optimization and deterministic calcula-
tion packages. Optimization approaches rely on iterative optimization algorithms to reduce
and solve a system of equations with a defined objective and constraints. This approach is
24
used when high variability exists in many decision variables but there is a defined objective
that should be reached such as a least-cost or maximum-return approach. For optimization
to provide meaningful results, the program must have many degrees of freedom in order to
have license for altering the decision variables in order to find the best solution. Contrary
to deterministic solutions to systems of equations, there must be more undefined variables
in the program than there are independent equations. Excessive constraints placed on
an optimization model gives rise to trivial solutions or an infeasible problem. Although
there are many types of optimization problem, three of the most common are linear pro-
gramming (LP), mixed-integer linear programming (MILP) and mixed-integer non-linear
programming (MINLP).
Linear programming methods rely on linearization techniques to convert non-linear ob-
jectives and constraints into linear ones which can be handled by the LP solver algorithm.
This type of optimization is the most widely-studied and forms the basis for other types
of optimization. The solution of an LP can be computed very quickly and even very large,
complex or inefficient programs can be solved using a modest amount of computing power.
MILP problems stem from the foundation of LP formulations but are allowed to include
integer and binary components. This type of program is a powerful addition to the LP for-
mulations as it allows programmers to include discrete decisions within the model. Many
complex problems can be reduced to MILP formulations by applying techniques developed
specifically for this purpose. The MILP can be applied to a wider array or problems than
an LP and the solution algorithms are similar to those used for LP programs. Generally,
the algorithms for MILP solutions consist of solving many LP sub-problems in order to
find a feasible solution within the MILP super-problem.
25
MINLP formulations are another step in complexity from the MILP but are becom-
ing increasingly popular with the rise of inexpensive computational power. Attempts to
force non-linear constraints into linearity have been a focus of study for many years as
the computational expense associated with solving MINLP problems was high. With the
increase in computing power, solutions to large MINLP problems are becoming possible
and more widely-used. Though still computationally-intensive, MINLP optimizations are
very powerful and can be used to model even more complex systems than the MILP. Con-
straints that cannot be linearized can be included in an MINLP model and thus this type
of optimization can be applied to increasingly complex systems. Algorithms for solving
MINLP formulations are not as well-studied as those for LP or MILP problems as they
are less common and considerably more complex problems to solve. The basis for finding
solutions is similar in relation to the MILP algorithms as the MILP algorithms are to the
LP solvers. Generally, the MINLP solver uses many linear and non-linear solutions with
feasible integer solutions encompassed within the larger MINLP in order to bound the
search space and use alternative methods for decisions on which sub-problem to solve in
the next iteration.
Deterministic calculations involve utilizing known quantities and relations to provide
the solution to a problem with known inputs. Most popular calculation packages are
examples of this type. Deterministic calculations rely on known quantities and will return
a solution that may not be optimal. These calculations are used in situations where more
variables are specific or well-defined in order to quantify values for an unknown.
2.4.1 Software Used
The packages for optimization discussed herein are ‘A Mathematical Programming Lan-
guage’ (AMPL) and ‘General Algebraic Modeling System’ (GAMS). AMPL was developed
26
at Bell Laboratories and its development is currently under the direction of AMPL Op-
timization LLC. AMPL is a popular package for optimization used in both industry and
academia for linear and non-linear problems with both continuous and integer variables.
AMPL supports a variety of architectures common to deterministic programming packages
but are omitted from many optimization softwares such as handling looping commands and
case-specific declarations.
GAMS is another optimization package used in this work and is applied for the analysis
of the eco-park scenarios. GAMS is created and supported by the GAMS Development
Corporation and is designed for modeling and solving complex, large-scale models. Typ-
ically, GAMS is used for linear, non-linear and mixed-integer optimization problems but
also includes the capability for other model types. GAMS incorporates many solvers cre-
ated by research institutions around the world to create a broad-based architecture that
can be applied to many realistic situations.
Deterministic calculation packages used for this work include many well-known tools
such as Microsoft Excel and MATLAB but also include SimaPro, a life-cycle assessment
(LCA) tool, and GREET, a tool for assessing impacts of vehicles and fuels. SimaPro is
developed by PRe Consultants and includes many LCA databases from around the world
in order to provide a comprehensive analysis of life-cycle impacts for many products and
processes.
GREET is a calculation tool based in Microsoft Excel which is developed by the US
Department of Energy at Argonne National Laboratory. GREET is a specific life-cycle
assessment tool for analyzing the impacts and emissions from vehicles and vehicle fuels.
There are two series’ of GREET developed to date, the first series assesses the impact of
27
vehicle fuels while the second series analyzes the impact of the physical vehicles including
the materials and energy of construction etc. For the analysis herein, only the first series
of GREET was utilized, specifically, version 1.8b.
2.5 Environmental Metrics
This research also contributes to society by developing a method for evaluating industrial
relationships and by assisting in the planning of new facilities. This will benefit citizens
as the optimization will take environmental factors into account and will attempt to mini-
mize the overall waste from facilities that could otherwise affect living conditions in areas
surrounding these facilities. The impacts on air, water and land can all be considered and
the importance of the environment is taken into account in addition to the contribution
to the economic performance of industrial processes. If construction of a new chemical
facility can be made more economically feasible by its integration with other processes in
the region, construction and factory workers would also be required to build and operate
these facilities. This would contribute to the economic stability in the region in addition
to causing a decline in unemployment rates. As such, a number of environmental metrics
will be used to evaluate the performance of the eco-park scenarios compared to similar
independently-operated facilities.
2.5.1 Climate Change
The threat of climate change from anthropogenic sources of greenhouse gases (GHGs) has
been a source of environmental and political debate for many years. Public demands on
policy-makers are forcing governments to consider energy supplies from more renewable
sources and that regulations be placed on companies who are contributing to an increase
in the CO2 content in the atmosphere, as CO2 emissions are the principle contributor to
28
the enhanced greenhouse effect contributing to climate change. As such, CO2 is used as
the reference metric for greenhouse gas emissions from any source. One of the benefits
of eco-parks is that efficiency in energy use leads to a reduction in GHG emissions. This
reduction is one of the focal points of recent proposals for eco-parks as it is a burgeoning
topic in the public realm of environmental stewardship.
2.5.2 Air Pollution
Air pollution has been a subject of concern for health officials in many major cities, in-
cluding Toronto, Ontario, Canada. One estimate of the number of fatalities due to air
pollution in Toronto claims that 1700 deaths per year are attributable to air pollution
[36]. The combination of light, oxides of nitrogen (NOx) and volatile organic compounds
(VOCs) produces ground-level ozone (e.g., photochemical smog) which is considered to be
a major contributor to air pollution and premature death. By reducing NOx emissions
and centralizing other emissions to a locale removed from urban centres, such as Toronto,
eco-parks can be a factor in reducing these fatalities.
2.5.3 Ozone Depletion
Although not particularly pertinent in this study, ozone depletion became a major concern
several decades ago as chloro-flouro-carbons (CFCs) were found to deplete the ozone in the
atmosphere, leading to an increase in ultraviolet light striking the surface of the Earth which
had potentially disastrous consequesnces for humans. Although ozone-depleting substances
are not manufactured in the proposed network, this is a common impact category in life-
cycle assessments to ensure that additional production of these substances is not incurred.
Ammonia is also being studied as a potential refrigerant as it does not deplete ozone and
is capable of operating within a refrigeration cycle which would offset the usage of CFCs.
29
2.5.4 Acid Rain
Oxides of sulphur and nitrogen (SOx and NOx, respectively) are considered to be the major
culprits behind acidification and acid rain. These emissions can be produced in a number
of chemical facilities but tend to be found in much larger quantities as emissions from
electricity generation stations, specifically fossil fuel plants. By capturing emissions from
the power generation in this network, it is expected that gases impacting acidification will
be reduced; additionally, biomass can be used as a fuel source for gasification and should
thus emit less SOx than would the equivalent electricity production from coal or oil.
2.5.5 Resource Conservation
One of the major impacts that an eco-park can have is the ability of this arrangement
to conserve resources. In the case of EINs, energy feedstocks are conserved by utilizing
heating and cooling efficiently but also by using alternative fuel sources to replace fossil
fuel energy. Other resources are conserved by appropriately using the products and co-
products of other eco-park processes instead of requiring extensive production, packaging
and shipping of feedstock materials.
2.5.6 Habitat Destruction and Fragmentation
Localizing many facilities in close proximity would reduce the amount of deforestation and
habitat destruction to be absorbed by local wildlife. An arrangement of disparate plants
and shipping routes would only endanger wildlife by fragmenting their habitats and may
lead to animal management issues as can be the case in many rural climates. This can be
seen as efficient land use and the economies of scale associated with land-clearing ventures
would also serve to reduce the capital required for start-up operations.
30
2.5.7 Solid Waste
Another central idea to the vision of eco-parks is efficiency in terms of material usage and
the percentage of a material that is used in final products. Although a traditional plant
may only utilize 70% of a feedstock in its product, eco-park collaboration allows the unused
portion to be integrated into other products or processes and may increase the material
utilization of the feedstock to almost 100%. Although solid waste is typically unavoidable,
the ratio of solid waste to material input can be greatly reduced by using other portions
of the feedstock for other applications [3].
2.6 Financial Metrics
While environmental metrics are becoming increasingly important to business leaders, com-
panies are still responsible to their shareholders to show solid and sustainable economic
performance. The use of a dual-objective function in this work allows for profitability
to also be considered in the optimization and leads to a solution that proves to be envi-
ronmentally responsible as well as being economically feasible. This research is intended
to reinvigorate discussions in the industrial sector regarding issues such as sustainability,
process symbiosis and collaborative efforts.
The eco-park concept is synonymous with polygeneration, industrial symbiosis and the
like. All of these terms are based upon the concept of a diverse group of industrial produc-
ers cooperating to achieve a common goal of cost-savings and/or reduced environmental
impact. The literature has also shown that developing these eco-parks can very much be
a driver for innovation and development of new technologies [37, 38]. The Dow Jones
Sustainability index, DJSI, is an indicator used to manage investment funds based on sus-
tainability metrics. The funds developed using this index have shown solid growth since
31
the inception of the program and is an indicator that sustainability within an organization
can have significant impacts on profitability. Though the index also focuses on business
practices and management styles, the overarching reality is that sustainability within a
company yields financial performance results [39, 40].
2.7 Summary
This chapter explains the context in which this thesis is completed. Economics and envi-
ronmental concerns are assessed in terms of life-cycle impacts within the domain of eco-
industrial integration. Current literature concerning eco-industrial integration is primarily
focused on either economic or environmental principles while neglecting the other; however,
this work is unique in its assessment of both concerns as part of the objective function in
an optimization model. In addition, quantifiable assessment of eco-industrial benefits is
scarce as the majority of the work to this point has been primarily qualitative and lacking
empirical support whereas this research is completely focused on the quantifiable benefits
of eco-industrial integration.
32
Chapter 3
Air quality and environmental
impacts of alternative vehicle
technologies in Ontario, Canada
Chapter 3 is based on the previously published work “Air quality and environmental im-
pacts of alternative vehicle technologies in Ontario, Canada” by Kantor et al. [41] as seen
in the International Journal of Hydrogen Energy 35(10):5145-5153 and is reproduced with
permission from the International Association of Hydrogen Energy. The thesis author’s
specific contributions to this paper were to develop the model of emission reduction poten-
tials, conduct the simulations, prepare the graphics and results, write the final manuscript
and respond to the comments of reviewers. This work was conducted with direction from
the project supervisors, Dr. M. Fowler and Dr. A Elkamel, who are co-authors on the pub-
lication. Amirhossein Hajimiragha contributed with primary modeling of the electricity
grid in the province of Ontario to determine the supportable penetration of alternative-fuel
vehicles.
33
3.1 Introduction
The economies of the developed world are increasingly expanding to include “green” tech-
nologies and processes that take into account the social, environmental and economic conse-
quences of business decisions. Western society, as a whole, is demanding that the products
and services that it uses are less harmful to human health and to the environment. The
transportation industry has made significant advances in fuel efficiency of the vehicle power
trains and reduction of emissions in the past decades, but more is expected from this sec-
tor. As the price of gasoline rose in combination with this societal green shift, vehicle
companies have commenced production of hybrid electric vehicles and other fuel-efficient
vehicle types. The impetus of this shift was to supply consumers with vehicles that would
decrease their ecological footprint as well as reduce the cost associated with purchasing
fuel. In recent years, energy security has also become a driving force for change in vehicle
fuel types. One of the societal concerns often overlooked is the impact of alternative-fuel
vehicle usage on the air quality in the urban environment. It is the purpose of this chapter
to assess the impact on air quality stemming from the operation of alternative-fuel vehicles
in urban environments.
While several studies have based the comparison of alternative fuel vehicles (AFVs) on
least-cost comparisons or other economic metrics [42, 43, 44, 45, 46], this study is purely
focused on air quality. The effects on overall air quality are considered with respect to
climate change potential and acidification. The special focus of this study is on urban air
quality as it can be of major concern in large centres of population.
This chapter is concentrated on the province of Ontario and specifically the city of
Toronto for two major reasons. The primary reason for this focal point is that Ontario
represents the most highly-populated province in Canada which naturally leads to a higher
34
level of concern from the increased number of individuals affected. The second reason is
that urban air quality in Toronto is specifically an area of concern due to the estimated
fatalities in this city. Traffic volumes in smaller cities would induce less concern as the
concentration of urban air pollutants is directly proportional to the emissions from vehicle
traffic. In addition, the data availability for Ontario in general and Toronto specifically is
more widely available due to the concerns mentioned above and to the increased govern-
ment resources attributed to gathering and analyzing this data.
The AFVs considered for this analysis are fuel cell vehicles (FCVs), plug-in hybrid
electric vehicles (PHEVs) and fuel cell plug-in hybrid electric vehicles (FCPHEVs). The
reason that these vehicle types were chosen is that they represent the most promising
technologies for partially replacing fossil fuels in conventional vehicles. The transition of
vehicle drive trains will begin with electrification of the vehicle drive train which allow for
hybridization with electric motors. Hybrid electric vehicles (HEVs) can make modest gains
in fuel efficiency mainly through the use of regenerative braking.
Once the drive train is completely electrified, the power train can be composed of a
combination of batteries and some type of range extender technology (e.g., gasoline, diesel
or fuel cell) to recharge the batteries onboard or provide electricity in parallel with the
batteries [47]. This differs from the methodology considered by Thomas [48] as this work
includes electric vehicles with range extenders and is not a comparison between FCVs and
battery-electric vehicles (BEVs) considered by Thomas [48]. FCV in this case refers to
compressed gaseous hydrogen as the technology is simpler and would likely be commer-
cialized before options that use liquefied hydrogen as the fuel.
The FCPHEV would operate as a normal plug-in vehicle except that the energy sup-
35
ply for the charge-sustaining mode would be supplied by hydrogen fuel cells and not by
gasoline. Other AFVs could also be compared on an emissions-per-distance basis but the
interest of this chapter is on the impact that could be achieved through greater utilization
of base-load electricity. As such, this study focuses on the near term transition technology
of the PHEV which will use electricity to recharge batteries, and the FCV which uses base-
load electricity to generate hydrogen. This analysis also represents the most promising near
term technology transition to PHEV and the technology with the greatest potential for
emissions reduction in the long term (FCV). The transition between the near-term adop-
tion of PHEVs to the eventual transition to FCVs is examined by Suppes [30].
Thomas [49] states that FCVs are the only vehicle technology that has potential to
virtually eliminate problems relating to urban air pollution. In this study, the effects of
vehicles on urban air pollution are considered in a similar fashion to the work of Thomas
[49]; however, the limitations of the Ontario’s electricity grid are incorporated into the
calculations. As such there are some notable differences in the environment, assumptions
and potentially the results. Specifically, the Ontario grid makes much less use of coal
as a generation source than the system Thomas [49] assumed, and greater use of nuclear
and renewable (mainly hydroelectric) sources. Also, this study assumes that only surplus,
base-load power is used for the transportation sector and thus represents a more feasible
transition scenario for the transportation sector, as the electricity is available and under-
utilized at this time.
The emissions from manufacturing the vehicles are not included as part of this study
at this time and will be considered in future analyses; however, these types of vehicles also
have increased emissions resulting from the manufacturing process would likely have less
impact on urban air emissions and are therefore likely to be insignificant for this study
where the main focal point is urban air quality. It should be noted that preliminary esti-
36
mates for the production of both types of AFV considered in this study show that current
production methods of traditional vehicle manufacturing emit less pollution and consume
less energy than current methods of AFV production. These preliminary results would
also be affected if centralized, large-scale production of AFVs were to exist on the same
level as traditional vehicle manufacturing.
Developing infrastructure has been considered by several authors for Southern Califor-
nia [42, 43, 50] with special focus again on the economics of its development. It is assumed
for this chapter that the distribution of hydrogen is available and thus the construction of
a distribution network is not included in the results of this study.
3.1.1 Health Effects
Toronto Public Health estimates that the number of annual deaths in Toronto from urban
air pollution is 1700 annually [51]. Estimates from the Ontario Medical Association (OMA)
[52] and Health Canada [53] estimates the number of fatalities is 5 800 throughout Ontario.
These deaths attributed to air pollution are most predominantly from lung diseases but
air pollution also partakes in increasing the rate of atherosclerosis which is a contributor
to heart disease and stroke [51].
The life cycle of hydrogen and its impacts have been studied previously [54, 55] in an
attempt to characterize the effect of hydrogen production in terms of life-cycle emissions
and sustainability. The use of hydrogen as a transportation fuel has also been considered
but comparisons between hydrogen and other transportation fuels are only now being de-
veloped [56, 57]. It is important to consider hydrogen as a transportation fuel relative to
other fuels in order to realize the consequences related to its mainstream adoption as a
37
transportation fuel. PHEVs have also been studied from a life-cycle perspective by several
authors [58] and this study is intended to compare these different types of AFV using a
realistic basis of penetration and adoption.
Overall and urban emissions from AFVs were both considered to be important since
overall emissions may affect climate change, acidification and other effects related to gener-
alized emissions into the air. Urban emissions were considered specifically for the purposes
of analyzing a possible decrease in fatalities caused by poor urban air quality. Photochem-
ical smog is particularly an issue when considering the large volumes of traffic that occur
during the rush-hour times in the greater Toronto area (GTA). Due to the location and
specifics of Toronto, smog formation is limited by the amount of nitrogen oxides (NOx)
present. According to the empirical kinetic modeling approach to photochemical smog, a
reduction in NOx would yield a much more pronounced effect on the reduction of photo-
chemical smog than would an even greater reduction in VOCs.
When considering urban air emissions, four major pollutants and one additional stres-
sor are considered. Two classifications of particulate matter, one having diameter less than
10 microns (PM10) and one of diameter less than 2.5 microns (PM2.5), are generally con-
sidered to be the most harmful to human health and are also the eventual products from
some other pollutants [59]. This small particulate matter is capable of penetrating deep
into the human lung, causing irritation and is too minute to be rejected by natural human
mechanisms [59]. VOCs and NOx react with sunlight to form photochemical smog which
is generally the largest contributor to urban air pollution in industrialized countries. Re-
ducing the synthesis of photochemical smog is a top priority for individuals involved with
addressing urban air quality in major cities. Athens, Greece and Beijing, China among
several other cities that have made similar laws, institution of bi-daily driving was initiated
in an attempt to partially curb the creation of photochemical smog. Sulfur oxides are the
38
remaining stressor and are generally viewed to be more of a significant factor with regard to
acidification than urban air pollution; nevertheless, it does contribute to producing aerosols
and particulate matter in the troposphere.
3.2 Modeling and Results
3.2.1 Data Gathering and usage
The current GHG emissions in Canada are shown in Figure 3.1 [36]. It is important to
note that these emissions are the overall emissions for Canada and are not specific to the
urban air quality which is considered to be of major concern due to the annual fatalities
exhibited in the GTA from air quality issues. For analysis of the impacts of AFVs, the
total emissions can be compared to the current overall emissions in Ontario. The results of
these comparisons can be realized as a percentage increase or decrease in each particular
emission type. For pollution that is mainly of concern in the urban setting, emission levels
are significantly harder to quantify due to the number of emission sites and the varied
locations of these sites as well as their relative severity.
The generation mix considered in this research is the approximate Ontario generation
mix shown in Figure 3.2. While this generation mix is expected to change, the relative
levels of production from each source should remain consistent. The reduction in emissions
in this study are calculated using this energy mix under normal conditions whereas the
base-load contribution is used in the circumstances that base-load power can be assumed
to be utilized (i.e., for hydrogen production). The current base-load generation mix is
shown in Figure 3.3.
39
Figure 3.1: Canadian greenhouse gas emissions by sector [36]
3.2.2 Methodology
The software packages of AMPL and GREET 1.8b [60] were used to complete the analysis
presented herein. The number of vehicles that can be feasibly supported by the current
electricity grid in Ontario including the planned modifications was found by modeling
the scenarios in AMPL. AMPL is a modeling language for mathematical programming
and is especially tuned for optimization scenarios. Every effort was taken to ensure ac-
curate results by using this model such as justifications of assumptions and sensitivity
analysis, the model is presented in [61] and a summary of the model is supplied in the
Appendix. The merits of this model are that it takes many factors into account such as
energy import/export, electricity prices, market penetration transition, generation capac-
ity, base-load generation mix, transmission capacity in addition to environmental credits
and vehicle data. It is important to note that the AMPL model uses conservative values
for predicting penetration levels based on information currently available concerning the
Ontario electricity grid. Conservative values are used in order to determine the smallest
possible number of AFVs that may penetrate the vehicle market in Ontario, Canada. In
reality, the penetration of AFVs that may be supported using Ontario’s energy grid are
expected to be larger than the results of the model indicate. Similar work has been com-
40
Figure 3.2: Overall electricity generation mix for Ontario [36]
pleted by Oi [44] for utilizing Japan’s base-load electricity for generating hydrogen. With
the resulting supportable penetration rates, GREET was used to calculate the pollution
abatement resulting from the adoption schemes.
The penetration rates for both FCVs and PHEVs were assumed to follow one of two
possible trajectories [61, 29]. These possible paths are shown below in Figure 3.4. The
first possible transition trajectory is labeled as such and yields a slow adoption and would
mimic the effects of an uncertain population who are hesitant to invest in a new technology
before it is proven. This transition rate has a slower initial response than the first transition
scenario but leads to a less volatile adoption scheme in which the general public steadily
gains confidence in the new technology. The second transition scheme presents a rapid
initial adoption of the AFVs which tapers off after the initial adoption phase before being
revitalized in the final years of the simulation. This transition scheme mimics a popula-
tion with environmentally and technologically oriented consumers who wish to incorporate
the new technology into their lives as soon as it is available. After the target consumers
41
Figure 3.3: Base-load electricity generation mix for Ontario [36]
have purchased these vehicles, a downturn in sales is experienced due to uncertainty in
the technology from the remainder of the general public [62]. As the technology is proven
and manufacturing becomes less expensive, members of the general population who were
previously hesitant are encouraged to purchase AFV technology which leads to the second
period of growth for this transition scenario. Only the results for the first transition sce-
nario are considered in this paper.
It is important to note that the scenario for the adoption of FCVs and PHEVs have
been compared as being mutually exclusive to illustrate the effects on the overall and urban
air pollution from adopting the individual vehicle types. This approach does not reflect a
realistic scenario given that new vehicle types will likely be adopted in parallel and none of
these will be sole type of AFV used, assuming that both were available. In all likelihood, a
combination of these vehicle types will be adopted as individuals make decisions based on
their own personal requirements. The emission changes from these reductions will then be
a combination of these vehicle types in quantities which could be estimated using consumer
surveys and adoption patterns of hybrid electric vehicles.
42
Figure 3.4: Transition scenarios for AFV penetration in Ontario
The FCPHEV has been included in this study in order to yield the maximum and
minimum emission reductions that can be supported in Ontario. It is important to note
that the AMPL model has not been attuned to produce the supportable penetration of
FCPHEVs and that the calculations for this vehicle type are based on the maximum sup-
portable penetrations for FCVs and PHEVs. Such analysis would be complex as not only
would electrical grid transmission constraints be considered, but the location, storage and
distribution of hydrogen needs to be considered as well. The potential for electrolysis to
provide voltage regulation within the electrical generation system would also positively af-
fect the use of available base-load power. The maximum achievable reduction in emissions
is calculated using the supportable penetration of PHEVs while the minimum is calculated
by using the penetration of FCVs. These estimates would lead to a power requirement
above the feasible limits of the planned Ontario grid or an underutilization of this grid for
43
the maximum and minimum cases, respectively. This methodology of developing a range of
market penetrations has previously been demonstrated for vehicle fleets as seen in research
completed by Wang, Ogden and Nicholas for the United States [63] and also by researchers
in Germany [64]. The benefit from this analysis is to be able to establish lower and upper
boundaries of emission reductions and not to predict actual emission reductions from the
supportable adoption of FCPHEVs.
For FCPHEVs, the calculations included data from the Canadian vehicle survey show-
ing that a daily drive for a vehicle is approximately 50 km [65] and additionally, 60% of
the distance driven can be powered by electricity (i.e., from the plug-in battery capac-
ity). FCPHEV energy usage will therefore consist of 60% grid electricity and 40% gaseous
hydrogen produced by electrolysis. The penetration rates for FCPHEVs are discussed in
more detail in the subsequent section 3.2.3.
3.2.3 Results of Supportable Penetration and Vehicle Growth
The population growth in Ontario and the percentage of Ontarians who currently own
vehicles can be used to predict the number of vehicles that will be present in Ontario in
future years. This information is found in Table 3.1. This table also shows the penetra-
tion rates of FCVs and PHEVs in Ontario for each given year based on the two transition
schemes addressed previously. The analysis was completed for two final penetration rates
of FCVs due to the fact that locating future generation projects in different regions have
a significant impact on the final supportable penetration.
A conservative estimate of 1.2% penetration of FCVs in Ontario is based upon new
nuclear generation capacity in the Bruce zone. If, instead, the location of this generation
44
is in the Toronto zone, the supportable penetration of FCVs climbs to 2.8% [29]. This is a
change which leads to further comparison of the two vehicle types. All further calculations
use only the estimate of almost 2.8% because it is the most probable scenario but similar
calculations have been completed for the alternative penetration rates. The correspond-
ing penetration of PHEVs is almost 6% [66]. One assumption in the calculation of these
penetration rates is that the social cost of carbon dioxide emissions is approximately $ 35
per tonne. This assumption is based on the work of Pearce [67] but was found to have an
almost-negligible impact on the penetration rates calculated by the model [61].
The drivable distance for PHEVs and FCV must be compared in order to be able to
compare their emissions on the same basis. The assumptions made at this stage are that
the all-electric operating range for PHEVs is 30 km per day (i.e., per overnight charge)
and that the annual mileage for a FCV is 20 000 km which corresponds to the approximate
annual mileage for a conventional vehicle. The drivable distance for these two vehicle types
can then be found and is shown in Table 3.2.
Penetration rates based on regional adoption would likely yield different results for air
quality as residents in urban areas and those who commute short distances on a frequent
basis may be more inclined to purchase AFVs than individuals having longer commutes or
living higher distances from urban areas. These speculations are not included as definitive
research is not available to confirm these market predictions.
As mentioned in the previous section, the number of FCPHEVs considered will be
equivalent to the number of PHEVs for the maximum-reduction case and will be equiv-
alent to the supportable number of FCVs for the minimum-reduction case. Though the
results from this analysis are either slightly high or low based on the planned developments
45
Table 3.1: Calculated number of vehicles in Ontario and penetration for FCVs and PHEVs
for both transition scenarios
Transition 1 Transition 2
Year Total Number of
Vehicles
in Ontario
(thousands)
Penetration
of FCVs
(%)
Penetration
of PHEVs
(%)
Penetration
of FCVs
(%)
Penetration
of PHEVs
(%)
2008 7074 0.00 0.00 0.00 0.00
2009 7155 0.04 0.09 0.04 0.08
2010 7237 0.07 0.16 0.20 0.44
2011 7321 0.12 0.25 0.45 0.96
2012 7405 0.18 0.38 0.69 1.48
2013 7491 0.27 0.58 0.92 1.97
2014 7577 0.39 0.84 1.11 2.37
2015 7665 0.54 1.15 1.23 2.64
2016 7755 0.72 1.55 1.32 2.82
2017 7845 0.93 1.99 1.37 2.93
2018 7937 1.17 2.50 1.41 3.02
2019 8030 1.41 3.02 1.46 3.12
2020 8124 1.67 3.58 1.54 3.29
2021 8219 1.93 4.15 1.67 3.58
2022 8316 2.17 4.66 1.86 3.98
2023 8414 2.41 5.16 2.12 4.55
2024 8514 2.61 5.58 2.43 5.21
2025 8615 2.80 6.00 2.80 6.00
46
Table 3.2: Calculated number of each type of PHEVs and FCVs in Ontario and the drivable
distance
Year Number of
FCVs
in Ontario
Number of
PHEVs
in Ontario
kms
drivable
by FCVs
Electric
kms
drivable by
PHEVs
2008 0 0 0 0
2009 2840 6090 5.680×107 6.665×107
2010 5370 11500 1.074×108 1.259×108
2011 8670 18600 1.734×108 2.034×108
2012 13200 28300 2.638×108 3.095×108
2013 20100 43100 4.021×108 4.717×108
2014 29600 63400 5.919×108 6.945×108
2015 41200 88300 8.237×108 9.664×108
2016 56200 120000 1.123×109 1.318×109
2017 72800 156000 1.455×109 1.708×109
2018 92600 198000 1.852×109 2.172×109
2019 113000 242000 2.260×109 2.651×109
2020 136000 291000 2.715×109 3.185×109
2021 159000 341000 3.181×109 3.731×109
2022 181000 388000 3.617×109 4.244×109
2023 203000 435000 4.056×109 4.758×109
2024 222000 475000 4.436×109 5.204×109
2025 241000 517000 4.824×109 5.660×109
47
to the Ontario electricity grid, the ultimate result is to establish the maximum and mini-
mum reduction of emissions. It should also be noted that in the case of FCPHEV there is
likely to be a wide range of fuel cell and battery combinations aboard vehicles, especially
during the transition phase as hydrogen distribution infrastructure is developed.
3.2.4 Pollution Abatement Results
The two scenarios presented here, which correspond to the adoption of FCVs or PHEVs,
must be considered as mutually exclusive. Both penetration rate assumptions depend on
maximum usage of the Ontario energy grid and thus cannot proceed in concert. As men-
tioned previously, other AFVs have not been considered as they could only be compared
on a per-kilometre basis which would not add significant value to this work.
The two areas of concern with respect to emissions are in the overall and urban sce-
narios. For comparison purposes, the emission reductions resulting from the adoption of
these AFVs are shown in Figures 3.5 – 3.11. Figure 3.5 represents the greenhouse gas
reduction and CO2 reduction simultaneously for consideration in the overall abatement of
gases that may contribute to global warming. Because these emissions are not suspected
to have appreciable effects on urban air quality, they are only analyzed from this overall
perspective. The other emissions are found to have effects in both the overall setting as
well as having an impact on urban air quality; therefore, the calculated reduction of a
particular emission is shown as an overall reduction as well as an urban reduction. Note
that these values are reductions, so a negative value is the product of an increase in that
emission.
Figure 3.5 Illustrates that the majority of the greenhouse gas emissions can be at-
48
Figure 3.5: GHG and CO2 reduction in Ontario
tributed to the release of carbon dioxide into the atmosphere which is illustrated by the
fact that the GHG and CO2 lines are almost identical. Since the transportation sector
in Ontario emits approximately 3.76 × 107 tonnes of GHGs per annum based on data
from 2005, the normalized reduction can also be calculated and is shown in Figure 3.6.
It is observed that the reduction in the transportation GHG emissions in Ontario would
reach the level of 3 to 3.5 percent by 2025. The FCPHEV predictions yield the maximum
and minimum reduction from the transportation sector. It is observed that the range of
reduction in GHG emissions from the transportation sector would be between 3% and 5.8%.
By analyzing the information in Figure 3.7, it is observed that PHEVs exhibit superior-
ity in reducing VOCs in both the overall and urban scenarios when compared to the FCV.
As the major concern with VOCs is related to photochemical smog production in urban
49
Figure 3.6: Normalized reduction in GHG emissions
areas, it is important to note that PHEVs show approximately 100 tonnes of reduction per
annum above the levels that can be achieved by FCVs. The FCPHEV scenarios yield the
maximum and minimum reduction of VOCs that could be achieved by adopting a fleet of
FCPHEVs that would follow the high penetration rate of PHEVs or the lower penetration
rate of FCVs. For the emission of VOCs, FCPHEVs could greatly exceed reductions from
either FCVs or PHEVs in the overall scenario but would only show a very slight benefit
in urban areas relative to PHEVs. As the most significant contribution of VOCs is urban
air pollution, PHEVs and FCPHEVs are approximately equivalent in terms of emissions
while they both yield a greater reduction than FCVs.
Figure 3.8 demonstrates that both PHEVs and FCVs will have similar effects on urban
air quality in terms of reduced NOx emissions. Due to the fact that the photochemical
50
Figure 3.7: VOC emission reductions in overall and urban settings for AFVs
smog reactions in a region such as Toronto are limited by the amount of NOx available
for reaction, the urban NOx reductions are of greater concern than similar levels of re-
duction in VOCs. As both PHEVs and FCVs have approximately the same reduction in
NOx, neither can be considered to be superior in terms of reducing this emission. The
other pertinent information to be observed from Figure 3.8 is that the FCVs demonstrate
a larger overall reduction of NOx which would lead to a slight decrease in acidification.
While this attribute is positive, this reduction in NOx is unlikely to decrease acidification
by an appreciable amount as the annual emissions of NOx and SOx in Ontario exceed this
by over four orders of magnitude [68].
The maximum and minimum reductions in NOx from adoption of FCPHEVs are also
shown in Figure 3.8. It is observed that NOx reduction by a maximum number of
51
Figure 3.8: NOx emission reductions in overall and urban settings for AFVs
FCPHEVs would reduce NOx emissions by approximately 50% more than the support-
able number of FCVs and more than double the reduction when compared to PHEVs
considering overall emission levels. Since the urban emission reductions are more pertinent
for NOx, it is important to note that at a maximum level of penetration, FCPHEVs could
reduce urban NOx emissions by an additional 75% over the levels that either PHEVs or
FCVs are able to achieve at their respective levels of supportable penetration. Additionally,
the reduction levels for a minimum adoption of FCPHEVs show only a slight deficiency in
NOx emission reductions when compared to PHEVs and FCVs for their respective sup-
portable penetration levels.
The reduction of particulate matter emissions with diameter less than 10 microns is
shown in Figure 3.9. The dominant trend in this figure is the increase in overall PM10
52
Figure 3.9: PM10 emission reductions in overall and urban settings for AFVs
related to the adoption of PHEVs. This increase is related to the increased particulate
emissions from burning coal using the current generation mix in Ontario. The negative con-
sequences of particulate emissions associated with coal electricity generation and PHEVs
use is confirmed by other studies [49]. The PM10 emissions from a maximum number of
FCPHEVs follow a trend similar to that shown by the PHEVs with only slightly lower
increases in emissions. It is observed from the figure that the overall emission of PM10
will increase in a range of 190-370 tonnes per annum by adopting FCPHEVs; however,
the urban reduction in this emission will be similar to the levels observed for FCVs and
PHEVs. By adopting the maximum amount of FCPHEVs, the reduction in PM10 could
achieve a reduction in PM10 emission of 75% greater than that achievable by PHEVs or
FCVs.
53
Figure 3.10: PM2.5 emission reductions in overall and urban settings for both types of
AFVs
Taking into consideration that the annual emissions of PM10 in Ontario exceeded 1.04
million tonnes in 1995 (including open sources) [68], an overall increase of 400 tonnes per
year by 2025 is hardly significant. The reduction of particulate matter in the urban air is
very similar for both FCVs and PHEVs and is 40-42 tonnes per annum in 2025.
The trends for PM2.5 have similarities to those found for PM10 as is exhibited in Figure
3.10. The overall increase in emissions of PM2.5 for PHEVs is significant as was the case
in Figure 3.9 for PM10 with the cause again being the increased use of coal for generating
the electricity for these vehicles. It is again observed that the emissions of PM2.5 in the
urban setting will decrease by approximately the same amount for both PHEV and FCV
technologies with reduction reaching 29 and 31 tonnes for FCVs and PHEVs, respectively.
54
The upper and lower reduction limits for urban PM2.5 by adopting FCPHEVs can be ob-
served to bound the reductions predicted for PHEVs and FCVs; the lower limit appearing
slightly below the reductions from FCVs and PHEVs while the upper reduction limit is
significantly above the reduction for the other two vehicle types.
The emissions of this size of particulate matter (PM2.5) are of most concern in the ur-
ban air as they have the greatest potential for harm to human health. It is observed from
Figure 3.10 that because of the hybrid nature of the FCPHEV, the range of overall emis-
sion increases from FCPHEVs falls within a small range of approximately 10 - 20 tonnes
per annum by 2025. As for PM10, though, the overall emissions are again overshadowed
by the province-wide emission of over 250 000 tonnes (including open sources) [68]. A
minor increase in the overall emissions is thus insignificant but reductions in the urban
environment could have slight positive impacts on population health.
From the two trends of particulate matter emissions, the data for SOx in Figure 3.11
is not surprising. The increased use of coal for electricity generation has again led to an
increase in emissions of SOx. SOx emissions are generally of concern when considering
acidification and thus the overall emissions are of particular interest. The SOx emissions
in Ontario in 1995 were estimated to be over 632 000 tonnes meaning that the increase of
790 tonnes annually by 2025 would be an increase of just over 0.12%.
The adoption of FCPHEVs shows an impact on overall emissions of SOx that encom-
passes a range of increasing SOx emissions from 320 - 620 tonnes per year by 2025. These
increases are slightly less than the increase attained from adoption of PHEVs yet are still
significantly greater than the actual reduction achieved by adoption of FCVs. The elec-
tricity requirement for the plug-in portion of the operational time would contribute to
55
Figure 3.11: SOx emission reductions in overall and urban settings for AFVs
increased emissions just as was the case for PHEVs.
SOx emissions in the urban setting are of considerably less concern than precursors
of photochemical smog and particulate matter and thus are not especially pertinent for
discussion here.
3.3 Summary
Using the electricity grid infrastructure and planned improvements allows for calculation
of the supportable penetration of PHEVs, FCVs and FCPHEVs in Ontario, Canada. From
this study, it is evident that a reduction in life-cycle emissions can be achieved by tran-
56
sitioning a fleet of conventional vehicles to alternative fuels. For the metropolitan centre
of Toronto, Ontario, alternative fuels of all types would decrease the urban emissions of
small particulate matter (PM2.5 and PM10) and NOx which are the major sources of urban
air quality problems. From a provincial or national level, each alternative vehicle type is
expected to decrease all emissions except for SOx, which are expected to increase slightly
in most situations.
The results show that FCVs and FCPHEVs are the most effective vehicle types for
reducing life-cycle emissions during operation; however, the refueling infrastructure for
these vehicles is not currently sufficient to make these vehicles practical. As such, it
is recommended that PHEVs should be adopted in the near term with a transition to
hydrogen as soon as the infrastructure allows.
The reduction in emissions on a national or provincial level from adoption of alternative
fuel vehicles is less significant than are the reductions in the urban setting and thus the
impetus for adoption of such vehicles from a policy standpoint would be to reduce health
impacts from urban air pollution.
57
Chapter 4
Optimized production of hydrogen in
an eco-park network accounting for
life-cycle emissions and profit
Chapter 4 is based on previously published work “Optimized production of hydrogen in an
eco-park network accounting for life-cycle emissions and profit” by Kantor et al. [69] as seen
in the International Journal of Hydrogen Energy 37(6):5347-5359 and is reproduced with
permission from the International Association of Hydrogen Energy. The thesis author’s
specific contributions to this paper were to develop the model, conduct the simulations,
prepare the graphics and results, write the final manuscript and respond to the comments
of reviewers. This work was conducted with direction from the project supervisors, Dr.
M.W. Fowler and Dr. A. Elkamel, who are co-authors on the publication.
58
4.1 Introduction
4.1.1 Problem Definition
The purpose of this chapter is to develop a method for optimizing the material and en-
ergy usage for an existing network of industrial facilities in order to reduce emissions and
waste generation, while optimizing material and energy output to maintain product output.
Specifically in this case, the eco-park is used to generate hydrogen for the hydrogen econ-
omy. The overarching goals are to reduce emissions and energy use without compromising
the process profitability. Through collaboration within an eco-industrial network or com-
munity of industrial facilities, chemical processors can reduce their environmental impact
while still pursuing profitability to maintain favour amongst shareholders. The quantita-
tive benefits of pursuing eco-park concepts within a network of facilities will be identified.
This will help to exhibit the possibilities for industry to collaborate in order to maintain
or increase profitability while reducing their individual impacts on the environment.
4.1.2 Eco-industrial Network Description
The concept of eco-industrial networks (EINs) has been discussed in detail in Section 2.1
and will not be discussed in detail here. The major principles behind EIN development
are to achieve environmental and economic goals by integrating the production and usage
of energy and materials from many facilities. This approach is the foundation of indus-
trial symbiosis to mimic the behaviour of the natural world. This chapter is focused on
developing a generalized linear program (LP) for optimization of these networks.
59
4.1.3 Analysis Methods
Economic and environmental objectives are both considered in this work in an attempt to
balance the interests of society with those of industry. Economics are easily measurable
based on construction and operation of chemical plants, purchase and sale of chemicals
and energy and other major-cost items. The environmental objectives in this work are
quantified by using llife-cycle assessment, as explained in Section 2.3.1 and are focused on
emissions from each facilitiy to the air, water and land.
4.1.4 Hydrogen Economy
In the context of integrated energy and production systems, different energy infrastructures
are to be studied simultaneously with the eco-park concept. In the last few years, the
concept of a hydrogen economy has attracted attention in industry and academia [70, 71,
28]. Hydrogen as an energy carrier can be produced from multiple energy resources like
fossil fuels, nuclear, and renewables for multiple end-uses; this has led to the development
of the hydrogen economy concept, which concentrates on the study of the economic aspects
associated with the production, distribution and utilization of hydrogen in energy systems
[29, 72]. Hydrogen is a desirable energy vector because it can be stored and used to generate
electricity. The use of hydrogen in transportation applications will result in decreased
urban air pollution and national greenhouse gas emissions, as well as diversified energy
production and security of energy supply [41]. Despite these benefits, in the present state
of technological development, there remains the need for development of a production,
distribution and storage network [71]. From the eco-park management point of view, the
use of hydrogen as an energy carrier is appealing, given its energy storage potential and
high value as an end product for the transportation sector. Hydrogen is both a product
and input in a variety of potential industrial facilities in an eco-park. A hydrogen economy
becomes an interesting possibility in the context of competitive electricity markets with
60
increasing amounts of intermittent renewable sources of energy (e.g., wind and solar) and
given the significant price differences between high and low demand hours for electricity, as
well as in urban environments where zero-emission vehicles are highly desirable [73]. Thus,
the use of hydrogen can address two key life-cycle metrics, the reduction of greenhouse
gases (GHGs) and the reduction of criteria air contaminants; specifically, urban smog-
generating emissions. Thus, if one considers in this market context that chemical and
energy generation plants are most efficient when operating at rated production and load
levels, the generation of hydrogen as a valuable end-product as well as energy storage
within the eco-park becomes highly desirable [74]. When the various advantages of the use
of hydrogen in transportation applications (i.e., in vehicles) are factored in, the importance
of studying the production, distribution and utilization of hydrogen in association with the
eco-park becomes evident. Thus, there is a need to consider hydrogen as an important part
of integrated eco-park systems. This chapter studies the production of hydrogen from an
eco-park perspective in association with the various hydrogen demands and uses with the
eco-park itself.
4.2 Network Description
4.2.1 Chemical Processing Plants
The network is comprised of several chemical production facilities including gasification,
CO2 capture, pressure-swing absorption, combined heat and power, as well as the manu-
facture of ammonia and urea. Mass and energy balances can be written for each network
node and are devised to maintain linearity in the model. Syngas generation can be carried
out utilizing a variety of fuels, j, that can be gasified in the corresponding set of gasifiers,
i. The variables are constructed of the form SpeciesUnitdirection to be interpreted that
CO2CHPout describes the amount of CO2 output from the CHP unit.
61
4.2.1.1 Gasification
The process of gasification typically consists of a hydrocarbon feedstock entering the pro-
cess where it is exposed to high temperatures, resulting in production of a mixture of
gaseous products referred to as syngas. The gaseous products, since the fuels are gener-
ally hydrocarbons, typically consist of carbon monoxide, carbon dioxide, water, hydrogen
gas and nitrogen products when air is used as the source of oxygen for the process. The
resultant gaseous mixture is typically dominated by hydrogen gas and carbon monoxide;
therefore, gasification utilizes hydrocarbon fuels as would standard combustion, yet the
products of the process are available for further use as chemical feedstocks and to service
the hydrogen economy.
Gasification is typically not practiced in the energy production sector as it adds an
additional step to the traditional combustion-centric approach without yielding noticeable
benefits. One benefit of gasification is in the versatility of the approach with regard to
potential feedstock [75]. Combustion boilers focus on one source of fuel as the design must
be catered to the normal operating parameters of the system. Gasification units can be de-
signed to accept a wider variety of fuels so that dependence on one type of fuel is no longer
a constraint on the unit. This also allows for the units to utilize biomass as a feedstock
to displace the usage of fossil fuels when such biomass is available [76]. Utilizing available
biomass for producing syngas can greatly reduce the overall usage of fossil fuels within the
network while also reducing the emissions associated with transportation by using biomass
generated in nearby agricultural facilities [77].
In this case, biomass was selected to enhance the usage of renewables within the overall
EIN. Modifying reaction parameters can also allow the producer to adjust the syngas ratio
depending on the downstream processes and process input [78]. The balances for this unit
62
can be written so as to maintain the appropriate amount of syngas as feedstock to the
downstream processes, as can be seen by equations 4.1 - 4.4 while the syngas is generated
according to biomass gasification described by Van Der Drift et al. [79].
CO2Gout = CO2CHPin + CO2PSAin (4.1)
N2Gout = N2CHPin +N2PSAin (4.2)
H2Gout = H2CHPin +H2PSAin (4.3)
COGout = COCHPin + COPSAin (4.4)
Equations 4.5 and 4.6 represent the total sulphur produced from the gasification section
and the sulphur produced from gasifier i, given the sulphur content of the fuel feeding the
gasifier and the flowrate of this fuel, respectively. Ash is quantified in a similar way as
shown in equations 4.7 and 4.8
S =N∑i=1
SGi(4.5)
SGi=∑j∈Ωi
ySGijFij ∀i (4.6)
where ySGijFij represents the fraction of sulphur produced by fuel j being fed to gasifier
i.
A =N∑i=1
AGi(4.7)
AGi=∑j∈Ωi
yAGijFij ∀i (4.8)
where yAGijFij represents the fraction of ash produced by fuel j being fed to gasifier i.
The following four equations (4.9 - 4.12) represent the syngas product from each of the
gasifiers that is fed into the syngas header.
COi =P − γWGS
n(4.9)
63
where γWGS represents the extent of the water-gas shift reaction during gasification.
H2i =P
n+m2− k
(4.10)
where P, n, m and k are functions of the specific biomass used as a fuel.
N2i ' 0 for biomass gasification (4.11)
CO2i = γWGS (4.12)
The total amount of each gas being supplied to the syngas header is then calculated
as the summation of each gas from the individual gasifiers. The total mix of gas in the
syngas header is then found by Equations 4.13 – 4.16.
CO =N∑i=1
COi (4.13)
N2 =N∑i=1
N2i (4.14)
H2 =N∑i=1
H2i (4.15)
CO2 =N∑i=1
CO2i (4.16)
Equation 4.17 represents the total amount of fuel that is fed into gasifier i while equation
4.18 yields the calculation of the total amount of fuel j that is used in gasification where
Ωi represents the set of fuels acceptable in gasifier i.
Fi =∑j∈Ωi
Fij (4.17)
Fj =∑i
Fij ∀j ∈ Ω (4.18)
64
Supply constraints on the operation of the gasification section are shown as equation 4.19
which limits the usage of each fuel to an upper limit of availability to the network. In this
case, only biomass is used in the gasification process.
Fj ≤ FUj (4.19)
An upper limit is placed on the fuel flowrate due to possibilities of limitation in supply
or desirability of a given energy feedstock. In the case of biomass, only a certain rate of
agricultural waste might be available for a given time of year; therefore, it is necessary to
include the availability of crop waste as a function of the season or month. For fossil fuel
feedstocks, it may not be desirable to utilize the maximum amount available, and thus this
constraint can also be used to fix an upper limit on usage of certain fuels.
4.2.1.2 CO2 Capture
Historically, carbon dioxide was considered to be a necessary by-product of electricity pro-
duction yet is now considered by many as a pollutant. Certainly, CO2 is a greenhouse gas
(GHG) and the principal contributor to climate change. Capturing CO2 from a gas stream
has been a focal point of research in the energy industry in an attempt to create “clean
coal” plants in which the CO2 would be captured from the stack and then disposed of in
a manner that does not follow the traditional approach of releasing it to the atmosphere.
The technologies developed to date generally consist of a recirculating medium used to
capture the CO2 and then a process to remove the CO2 from the capture medium.
Captured CO2 can be purified and used for a wide variety of processes in order to avoid
emitting it to the atmosphere which would have little benefit over simply combusting the
coal to produce electricity [80]. Monoethanolamine (MEA) is one of the capture media be-
ing pursued for its high capacity for capturing CO2 from flue gas. An alternative medium
65
is ammonia, a common chemical product that has demonstrated various advantages in
its ability to capture CO2 [81, 82]. Ammonia is being used in the proposed network as
a product from the EIN and also as a precursor to the production of urea. By using
ammonia as a medium for CO2 capture, the need to import MEA is removed and the car-
bon capture process can be maintained by using make-up ammonia from the nearby facility.
The CO2 generated must be sufficient to supply all of the processes requiring it while
any remainder is sequestered. The balances on this unit are seen in equations 4.20 - 4.22.
CO2CCout = CO2GHin + CO2ECin + CO2MEin + CO2Uin + CO2SQin (4.20)
N2CCout = N2CHPout (4.21)
NH3CCin = 0.01SCO2 CO2CCin (4.22)
where SCO2 represents the solubility of CO2 in ammonia based on the operating param-
eters of the unit. 1% make-up of ammonia is used as a design rule to avoid build-up of
contaminants and deactivation of the ammonia [83, 84], which is described in Eq. 4.22.
4.2.1.3 Pressure-swing Adsorption
Pressure-swing adsorption (PSA) has been used for many years and is an industrially
mature process. PSA is used to separate one or more gas species from a mixture and can
be applied to a wide variety of gas streams as the adsorbent material may be varied to
suit the specific application. The inlet gas is passed over an adsorbent which attracts the
desired gas or an impurity in the stream. The remainder of the feed thus continues to the
outlet for release or further processing. This process will be used to separate hydrogen
from a gaseous mixture, and as such, PSA is a key technology for implementation within
the eco-park to support the development of a hydrogen economy which will demand a pure
66
stream of hydrogen for use in fuel cells. Additionally, this process is frequently found in
refineries and ammonia plants as hydrogen is required as a feed for some units in these
facilities. Thus, hydrogen produced from this section of the network will supply the other
facilities in the eco-park and may also provide hydrogen as a valuable co-product to the
transportation market [85, 86]. Equations 4.23 - 4.25 show the balances on the PSA unit.
COPSAout = COPSAin (4.23)
N2PSAout = N2PSAin (4.24)
H2PSAout = H2AMin +H2M (4.25)
Hydrogen product gases can be exported from gasification of hydrocarbon feedstocks fol-
lowed by purification of the gas streams using PSA. Alternative methods have also been
proposed to utilize off-peak electricity generation to electrolyze water for the production
of hydrogen. Efforts behind these initiatives to produce hydrogen gas are to stimulate
low-cost hydrogen for use in vehicles and to develop the ‘Hydrogen Economy’, termed as
such due to the concept being is that hydrogen is used as an energy carrier for powering
society. Transportation of people and goods within Canada represents 27% of the total
GHG emissions within the country [87]. Utilizing hydrogen as an energy carrier to power
commuter transportation could reduce these emissions by 3-6% by 2025 based only upon
the current infrastructure and planned improvements [41]. Thus, this reference [41] will
be used to develop the target hydrogen output from the eco-park. In addition, the life
cycle economic cost of hydrogen vehicles can be reduced by producing hydrogen in a more
efficient manner and also by economies of scale associated with producing the vehicle itself
[88, 50].
As mentioned previously, excess off-peak electrical generation can be used to electrolyze
67
water. The hydrogen produced from this process can then be stored for later use in a
fuel cell to generate electricity during peak hours. This would allow for additional peak
generation capacity without constructing additional peak- or base-load plants, leading to
a much more cost-effective energy system.
4.2.1.4 Ammonia Manufacture
Production of ammonia is another mature process that has been developed to produce
fertilizers for the growing agricultural sector. The feedstock for this process is typically
natural gas, which is processed to remove sulphur compounds and then reformed to produce
syngas. Upon separating the gases and introducing air as a source of nitrogen, ammonia
can be produced in large quantities. The common modern methodology for producing am-
monia is the Haber-Bosch process, converting nitrogen and hydrogen directly to anhydrous
ammonia in reaction 4.26.
3H2 +N2 → 2NH3 (4.26)
Ammonia will be produced in the proposed eco-park using a similar methodology although
the typical process will be much simpler as the feed gases are already free from sulphur
compounds and other impurities. The process may also accept unreacted ammonia from
the urea plant, depending on process conditions and geographical locations. LCA on
ammonia processing can provide details on improvements based on the modifications [89].
The governing equation for the ammonia section is shown by equation 4.27
NH3AMout = NH3CCin +NH3Uin +NH3M (4.27)
4.2.1.5 Urea Manufacture
Urea is primarily produced as a nitrogenous fertilizer for the agricultural industry but cur-
rently contributes to GHG emissions in the forms of CO2 and oxides of nitrogen in addition
68
to the heat and power required to operate the facilities [90]. The solid urea is broken down
into two ammonia groups and one molecule of carbon dioxide. Although several small-scale
methods have been developed for manufacturing urea, large-scale manufacturing methods
consist of combining the afore-mentioned two molecules of ammonia with one molecule of
carbon dioxide according to the Bosch-Meiser process in reaction 4.28.
2NH3 + CO2 ↔ (NH2)2CO +H2O (4.28)
In the proposed eco-park, carbon dioxide is readily available as a pure process stream
from the carbon-capture process and ammonia is also produced as a park co-product. One
advantage at this point is that the inefficiency related to compression of CO2 into dry ice
for transportation is not necessary. Since CO2 and ammonia can both be obtained from the
eco-park, the manufacture of urea may be one of the most profitable nodes of this process.
Environmentally, urea acts as a convenient transportation medium for urea and carbon
dioxide. The fertilizer pellets are a much easier method for applying ammonia to fields and
is a convenient source of carbon dioxide immediately available to growing biomass.
The production of urea from ammonia and carbon dioxide is a two-step reaction in
which the stoichiometric ratio of NH3:CO2 is 2:1. The reaction transpires according to
equations 4.29 and 4.30, shown below.
CO2 + 2NH3 ⇐⇒ NH2COONH4 (4.29)
NH2COONH4 ⇐⇒ NH2CONH2 +H2O (4.30)
Thus it can be seen that production of urea consumes ammonia and CO2 while producing
water as a co-product. Upon separation, the water can be recycled to be used in other
locations in the network. Because this is an equilibrium reaction, excess CO2 can be added
to the reaction to facilitate an equilibrium shift toward the production of urea. Analyses
have shown that an equilibrium conversion of 85% can be achieved by having excess CO2
available for reaction. The remaining reactants can then be recycled within the plant or
69
elsewhere in the network. For modeling purposes, 85% equilibrium conversion of ammonia
with 50% excess CO2 is assumed. The remaining reactants are available for recycle and
use elsewhere in the network. The balances on the urea section are shown in equations
4.31 - 4.34.
UUout = NH3Uin
(K
SR
)(4.31)
where K represents the equilibrium conversion within the reactor and SR represents the
stoichiometric ratio for the reaction.
CO2Uin = 1.5NH3Uin (4.32)
With the factor of 1.5 built in to the function so that CO2 is supplied in 50% excess of the
stoichiometric requirement.
NH3Uout = NH3Uin − 2UUout (4.33)
CO2Uout = CO2Uin − UUout (4.34)
4.2.1.6 Combined Heat and Power
Plants for generating heat and power, also termed cogeneration plants continue to receive
attention as the efficiency of these plants in producing electricity and useful heat exhibits
that they have potential for becoming a valuable part of the energy solution. The technol-
ogy is mature yet there are relatively few of these plants that have been built due to their
increased technical complexity and previously-undervalued ability to produce heat for use
in facilities or as district heating. The CHP node in the proposed eco-park is an important
part of the process as it provides heating for the facilities as well as an opportunity to
produce the electricity for the eco-park, further reducing the operational costs. This plant
will have the capability of completely oxidizing any residual carbon monoxide from the
syngas as well as hydrogen gas. The flue gas is to be treated by the carbon-capture process
70
so as to avoid emissions of CO2 to the atmosphere. The energy generation is a function of
the heating value of each of the gases fed into the unit. The heat generated for export is
either the heat required by other processes or the waste heat from the CHP process that
cannot be used elsewhere. The energy and heat generation are a split of the total energy
generation term, given that each has a different efficiency. The overall balances on the
CHP unit are shown in equations 4.35 - 4.40.
EG = (COCHPin) (LHV CO) + (H2CHPin) (LHVH2) (4.35)
EG represents the energy generated from the fuels fed into the CHP unit and LHV for
each gas is the lower heating value for each of the species.
HGX = EG (HeatSplit) (ηHG) (4.36)
EGX = EG (ElecSplit) (ηEG) (4.37)
where EGX and HGX represent the electricity and heat generated for export, respectively.
Similarly HeatSplit and ElecSplit represent the energy split between heat and electricity gen-
eration. In addition,ηEG and ηHG from these equations represent the efficiency of electricity
and heat production, respectively.
CO2CHPout = COCHPin + COPSAout + CO2PSAout (4.38)
H2OCHPout = H2CHPin (4.39)
N2CHPout = N2PSAout
(ηN2sep
)+N2CHPin (4.40)
While ηN2seprepresents the efficiency at which N2 can be separated.
71
4.3 Methodology
4.3.1 Introduction
The first steps to develop this model are to identify the nodes that are available to exchange
quantities of material and energy. The nodes have been identified in section 4.2.1 and will
be referred to by the corresponding indices for the remainder of this document:
• syngas generation G;
• pressure-swing adsorption PSA;
• combined heat and power CHP;
• carbon dioxide capture CC;
• ammonia production AM; and,
• urea production U;
With this list of facilities, it is possible to draw connections of products, co-products and
energy among the facilities. The objective function of the optimization is defined by the
development of the metric system shown in section 2.5. The objective function can be
manipulated to fit a wide array of scenarios including environmental indices as well as
profitability. The proposed network and connections is shown pictorially in Figure 4.1.
4.3.2 Definition of Environmental Metrics
Metrics for this optimization need to encompass all potential results of this collaborative
effort; therefore, the objective function must be formulated with these metrics in order
72
Figure 4.1: Proposed network
to achieve both goals of environmental responsibility and economic sensibility. The ques-
tion of what should be included in such an index is postulated by Ziegler [91] although
there are several options that already exist. An overview of different indexing systems
has been completed by several authors [92, 93, 20], yet a clearly superior index has not
emerged. Thus, it is important to review the applicable systems in order to develop an
appropriate set of metrics, and hence, an applicable objective function for the optimization.
Methods such as the Analytic Hierarchy Process (AHP) [94, 95], developed over 30
years ago are too primitive to be applied to this type of optimization although this is one
of the first standardized methods of analytical decision analysis for this type of problem.
Another system, called the Sustainable Process Index (SPI) was proposed in 1995 as a uni-
73
versally applicable index focused on sustainability. The SPI bases the sustainability of any
process on a ratio of area required for production to the area of consumption [96]. While
the pursuit of a single resultant output is beneficial for building optimization routines, it
is not particularly appropriate for the type of analysis considered in this work.
The Dow Jones Sustainability Index (DJSI) has been used for several years to base
investment decisions on the sustainability of a company as measured using financial, envi-
ronmental and social indicators [40]. This system has been used extensively in the financial
community to record extensive growth in investments. Unfortunately, this system relies
too heavily on qualitative information to be applicable in a numerical optimization algo-
rithm. Other metrics such as the waste reduction (WAR) algorithm have more empirical
clout but only apply to the extent that waste is reduced within a system [97, 98, 99]. This
methodology can be adapted and applied to be a part of the objective function but clearly
cannot be the only route pursued as it fails to include metrics traditionally important to
industry. Other indexes such as the Environmental Protection Index (EPI) [100] are also
somewhat applicable to this end, although it also excludes any mention of economic ben-
efits.
Emergy, exergy and e-green analyses have also been developed as an attempt to use
these systems to quantify benefits from eco-park networking [101]. These techniques are
very robust in their applicability but tend to focus more on energy and the efficiency of
energy usage within a process system. This type of analysis could be very effective in
monitoring or optimizing an energy-based eco-park but are unwieldy for application in a
material and energy exchange eco-park environment.
As none of the systems mentioned above are completely adequate for the analysis of an
74
eco-park, it is required to produce a new index which will account for both environmental
management and economic profitability. This index can then be used to formulate the ob-
jective function for an eco-park optimization. This metric basically consists of two parts,
one for the calculation of cost savings and the other for assessing the reduction in waste
and emissions. The caveat for this index is that there must be a comparable process so that
the difference between the two options can be calculated. The mathematical formulation
is shown in section 4.3.5.
4.3.3 Problem Formulation
The problem in this case is akin to a transportation/networking problem in a typical fashion
yet with several additional complexities. The first of these differences is that the eco-park
network includes many chemical reactions, which are atypical of a transportation problem.
Generally, a transportation or networking problem may have one or several goods/signals
transferred between nodes, yet the item in question remains unchanged. Chemical reac-
tions allow for a change in the good at each node as it may be converted into another
chemical and also be energetically altered.
Additionally, several types of good are being transferred and may not necessarily be
permitted to utilize the same transportation pathways. For example, although water or
natural gas may be transmitted through a pipeline at a capacity determined by the pipeline
infrastructure, electricity cannot be transmitted in a similar fashion. Thus it is required
that a minimum of two (electricity, materials) transportation pathways be implemented in
order to conduct goods between the facilities. Heat integration of the network plants could
potentially be transported using similar methods as the materials but at the extent that
the facilities are to be integrated, it is likely that heating must also be a separate pathway.
75
Similarly, some materials may require different forms of transportation than others and
thus the pathways for each chemical must be considered.
Simulations of eco-park concepts have been previously documented [102, 103, 104, 105]
but tend to apply non-optimal algorithms or metrics which are not based on life-cycle
thinking. The work presented here is an optimization model that uses a dual-objective
function in order to maintain profitability while reducing environmental impact based on
LCA metrics. This method is realistic as it does not compromise profitability for reduced
emissions but will yield a scenario amicable to both industry and society.
4.3.4 Selection of Optimization Package
The concept of combining chemical facilities into an eco-park system in an optimized way
would only be suited for an optimization package as it is desired to find the operating point
at which the profits and waste reduction benefits are maximized. GAMS software was se-
lected to complete the optimization as it is a powerful software language for optimization
and is well-suited to this type of problem [106]. Deterministic, numerical packages do not
provide the solution routines and are not specifically designed for optimization and thus
are not considered for use here.
GAMS stands for General Algebraic Modeling System, a commercial optimization pack-
age which is used extensively in both the academic and commercial realms for solving
optimization problems. The GAMS software employs a variety of strategies and solvers in
order to obtain the optimal solution for a given problem. The solver that is employed for
this model is CPLEX, which uses Simplex and barrier techniques for solving problems of
this type.
76
4.3.4.1 Transportation
Transportation of materials, heat and electricity are discussed as part of the problem
formulation in section 4.3.3 yet the aspects discussed were transportation associated with
the number of pathways required for goods to flow through the network. The environmental
impact of transportation has been studied for the consumer market and typically focuses on
greenhouse gases and total energy used per kilometre travelled. The transportation options
here, as the network has not yet been constructed, can be varied in order to reduce the cost
and emissions from the transportation of goods between facilities. Several environmental
factors have been shown to stem from altering the fuels used in vehicles [41] and vehicles
powered by electricity and hydrogen would integrate very easily into the network as these
two commodities are already produced.
4.3.5 Objective Function
The objective function for this optimization is a construct of two objective functions. The
two factors considered in this analysis are emission deviations from stand-alone plants as
well as economic incentives. It is important to consider both of these objectives so as not
to bias the output to be purely profit-motivated nor purely attuned to societal benefit from
reducing emissions. The portion of the objective function that governs the reduction in
emissions will tend to minimize the magnitude of all facilities; therefore, relying only upon
this metric, the plant sizes would be reduced to zero. In the scenario considered for this
work, note that the eco-park was constrained to provide hydrogen for 1000 fuel cell vehicles,
and the plants were sized accordingly as dictated by the optimal scenario. The economic
portion of the objective function is then incorporated to add realism to the optimization as
well as ensuring that the optimization will terminate with some plants have a size greater
than zero. Equations 4.41 - 4.43 show the condensed form of the objective function.
Z = WLCEJLCE +WeJe (4.41)
77
JLCE =
np∑p=1
EnvCostp (Sp − Ip) (4.42)
Je =
np∑p=1
[(ACC +OC)S − (ACC +OC)I ]p (4.43)
Where:
• JLCE = portion of the objective function attributed to the reduction in life cycle
emissions;
• Je = portion of the objective function attributed to the economics of the network;
• EnvCost = the environmental cost associated with a particular emission;
• p = representative of the particular chemical plant;
• np = the total number of plants;
• W = weighting factor for the economics (We) or life cycle emissions (WLCE);
• S=stand-alone facilities;
• I =integrated scheme;
• ACC =annualized capital cost; and,
• OC =operating Cost.
4.4 Results of Reduced Case
4.4.1 Reduced Case Model
In order to test the eco-park optimization theory and the objective function, five nodes were
extracted from the large case and the model was simplified to the one shown in Figure 4.2.
78
Figure 4.2: Reduced network configuration representative of the reduced case
The facilities accepted into the reduced case were carbon capture (CC), combined heat
and power (CHP), ammonia production (AM), urea production (U) and pressure-swing
adsorption (PSA). Transportation distances and types were removed from the model along
with their associated costs for simplicity, which basically assumes that the facilities will be
co-located within an eco-park complex. Additionally, the plants were able to scale linearly
instead of by discretized advances. These simplifications would result in a model of these
five nodes as if they shared a small geographical area with equipment that is custom-built
without incurring additional costs for such equipment.
4.4.2 Results
Simulating the five nodes mentioned above yields a result that can be shown in Figure 4.2.
Testing the case with the parameters presented in Table 4.1 yields the results shown in
Figure 4.3. The results from this trial represent a base case from which to experiment to
ensure that the model behaves logically in accordance with the constraints.
79
Table 4.1: Table of input parameters for basic case in reduced model
Parameter Description Value
We Optimization weight associ-
ated with economic perfor-
mance
0.5
WLCE Optimization weight asso-
ciated with life-cycle emis-
sions
0.5
Environmental Costs Description Value
CostCO2 Environmental cost/weight
of CO2
0.6
CostSOx Environmental cost/weight
of SOx emissions
0.05
CostNOx Environmental cost/weight
of NOx emissions
0.2
CostSW Environmental cost/weight
of Solid waste
0.15
4.4.3 Hydrogen Optimization
The network is comprised of several chemical production facilities including gasification,
CO2 capture, pressure-swing absorption, combined heat and power, as well as manufac-
turing of ammonia and urea. The amount of hydrogen produced from the network is fixed
to supply hydrogen 1000 vehicles in Ontario and the remaining plants are fixed in order to
accommodate the hydrogen production while maintaining an optimum value for the objec-
tive function. Assuming a 70 km kg−1 fuel efficiency for a hydrogen vehicle and 20 000 km
80
Figure 4.3: Reduced configuration representative of the reduced eco-park
a−1, the resulting mass of hydrogen for 1000 vehicles for one year of driving is 285 700 kg H2.
Varying the economic weighting between 0 and 1 by increments of 0.1 allows for the
stability of the optimization to be seen in Figure 4.4.
Although the overall reduction in emissions changes only slightly in the range of We
between 0 and 0.6, the plant sizes vary to accommodate the changes in the economic
weighting of the objective function. In the range between 0 and 0.3, carbon sequestration
is utilized in order to reduce emissions by a maximum amount, as the economic weighting
is relatively low and sequestering the CO2 would thus have little impact on the objective
function. In addition, production of urea during this stage is minimal as the economic
weighting is not at a level which would dictate that this production is necessary. When
the economic weighting reaches 0.4, it is sufficient to force the urea plant to increase in size
as a destination for the captured CO2 as the scenario is more profitable than sequestering
the CO2 resulting from gasification.
81
Figure 4.4: Results of producing hydrogen for 1000 fuel cell vehicles from [41]
At an economic weighting of 0.7, the optimization again reaches a new optimum value
for the objective function as the profitability is increasingly important. The optimization
dictates that the optimal solution for a high value of economic weighting includes reducing
the size of the carbon capture and urea manufacturing plants in addition to the combined
heat and power facility. In this case, the objective function is maximized by importing the
required electricity as it will reduce emissions less but is more profitable for the operation
of the remainder of the facilities.
The feasible region of the optimization is contained by several constraints and the
limiting constraint changes at the points mentioned above, i.e., when We=0.4, 0.7. The
differences take place in the size of the producing facilities for electricity, carbon seques-
tration and urea production. The hydrogen production is fixed at a level to power 1000
average hydrogen vehicles in Ontario, Canada, so the size of this overall eco-park facility
82
Figure 4.5: Normalized plant exports with hydrogen production for 1000 cars for various
values life-cycle emissions weighting
is variable yet the net hydrogen output is fixed.
Analyzing the plant sizes for a variety of values of WLCE yields the results in Figure
4.5. Positive values indicate an export of that product from the boundaries of the eco-park
while negative values indicate an import into these boundaries.
This figure shows the changing plant sizes for a variety of values of WLCE. The values
for each plant are normalized to their respective maxima to exhibit the sensitivity of
the model to the environmental and economic weighting factors. When We takes large
values, simulating a case in which profitability is the major concern, a large amount of
electricity is imported to produce the required hydrogen. Ammonia is also produced in
large quantities as it is a profitable product in this network. For scenarios of moderate We
83
values, environmental impact having a larger weighting relative to profit, a greater amount
of biomass is imported in order to manufacture the network products. The advantage in
emission reductions provokes the optimization to produce electricity, ammonia, urea and
heat in addition to the requirement for hydrogen. At WLCE = 1, ammonia production is
reduced to order to increase the production of heat and electricity while maintaining the
required amount of hydrogen. Although ammonia is a profitable product, the economics
of the scenario are unimportant as the weighting is entirely placed on the reduction in
life-cycle emissions from the network.
4.4.4 Greenhouse Gas Emissions Compared to Criteria Air Con-
taminants
Criteria air contaminants are defined by Environment Canada as pollutants that “cause
to air issues such as smog and acid rain. They are produced in varying quantities by
a number of sources, including the burning of fossil fuels” [107]. Varying the relative
importance of greenhouse gas emissions and criteria air contaminants leads to a result
showing some interesting results. Figure 4.6 shows that with a reduced weighting of GHG
emissions, the overall emissions from the network actual increase, as does the output of
urea and ammonia. Economics, in this case, drive the size of the plants and the emission
reductions follow as a result. As the network was developed for producing useful products
while focusing on reducing GHG emissions, it is not surprising that emissions may increase
with a very low GHG emission weighting. Until a GHG weighting of 0.3 is utilized, the
overall emissions from the network increase over the baseline case of no integration among
facilities. This is a reflection of the design complexities associated with the network plants
compared to traditional plants. These additional complexities contribute to an increase in
emissions as reducing GHG emissions yield no benefit to the objective value.
84
Figure 4.6: Analysis of the impact of varying the weighting of greenhouse gas emissions
and criteria air contaminants with life-cycle emissions weighting of 0.5 and profit weighting
of 0.5
At this time, the analysis does not account for the added benefit of avoiding the crite-
ria contaminants being generated in the urban environment from the burning of gasoline
(which is now offset by zero-emissions hydrogen), which would have a strong impact on ur-
ban health. Note the trends shown in Figure 4.4 would remain the same, only the amounts
would decrease as the number of vehicles in this scenario analysis is fixed.
Figure 4.6 also shows linear relations in each of these results which is to be expected
from the linear nature of the optimization program. As the relative importance of GHG
emissions is increased, the overall emissions are reduced as the network is designed for this
purpose. The LP model indicates that profit remains relatively unaffected compared to
the reduction in emissions and the objective value also corresponds to show the optimum
value changing with the reduction in emissions.
85
4.5 Summary
This chapter summarizes a method for constructing an optimization model to quantita-
tively assess the economic and environmental performance of a set of chemical facilities
utilizing a dual-objective function. It is evident from the results of this work that inte-
gration of chemical facilities into an EIN format yields both economic and environmental
benefits.
The products exported from each participant in the EIN varies with the weighting
placed on the two parts of the objective function and leads to four stable, optimal solu-
tions from a weighting of purely economic to purely environmental. The balance between
these objectives is important for policy-makers in communication with facility owners to
negotiate for the best option for profitability and also for reduced environmental impact.
The framework developed in this research allows for additional applications of the method-
ology for policy-makers across a variety of industries.
The sensitivity of the model through a range of economic and environmental weighting
factors shows that there are four optimal solutions in which the export of products from
each facility varies according to these two factors. The sensitivity of each of the individual
emission weighting factors is also considered and the results show that the objective value
and the emissions portion of the objective function vary linearly with changes in these
individual weighting factors.
The results show that emissions from the proposed network can be reduced and that the
profitability of the network is also maintained. Biomass gasification is used as a feedstock
to the network and an upper limit is placed on its availability in order to maintain the
sustainability of the EIN. The optimization also operates under a minimum constraint
86
of hydrogen production for 1000 fuel cell vehicles. This chapter is representative of the
‘cradle-to-gate’ emissions from the participating industries whereas the work in the previous
chapter exhibits the benefits from the entire ‘cradle-to-grave’ utilization of hydrogen as a
transportation fuel.
87
Chapter 5
Optimization of Material and Energy
Exchange in an Eco-park Network
Considering Three Fuel Sources
Chapter 5 is based on the forthcoming work “Optimization of Material and Energy Ex-
change in an Eco-park Network Considering Three Fuel Sources” by Kantor et al. [108],
in press with the International Journal of Advanced Operations Management and is repro-
duced with permission from the International Journal of Advanced Operations Manage-
ment. The thesis author’s specific contributions to this paper were to develop the model,
conduct the simulations, prepare the graphics and results, write the final manuscript and
respond to the comments of reviewers. This work was conducted with direction from the
project supervisors, Dr. M.W. Fowler and Dr. A. Elkamel, who are co-authors on the
publication.
88
5.1 Introduction
The goal of this research is to further the efforts made in exploring eco-industrial networks
(EINs) as a form of increased material and energy efficiency for manufacturing a given set
of products. The benefits of operating a facility within an EIN where inputs and outputs
are exchanged must be quantified in order to encourage development of these networks.
As such, this work contributes to empirically modeling an EIN with emphasis on reducing
environmental impacts while maintaining or improving profitability of each facility within
the network. Life-cycle assessment (LCA) has been utilized in order to assess the environ-
mental impact of the network facilities while profitability is computed using market pricing
for the exports from the eco-park network. The baseline production for the network is to
provide hydrogen for 1000 fuel cell vehicles which are intended to also decrease the emis-
sions burden from the transportation sector as shown by Kantor et al. to be beneficial
for urban air pollution and overall emissions [41]. Commercialization of hydrogen fuel cell
vehicles is expected from most major vehicle manufacturers in 2015, and thus this research
also contributes to the consideration of the transition to the ‘hydrogen economy’ [109].
The concepts of operating facilities within an EIP arrangement are documented in the
literature. The benefits most often touted are increased efficiencies in the form of:
• reduced energy intensity through energy exchange and heat integration;
• reduction in transportation costs/impacts;
• reduced material waste;
• increased profits;
• reduced water use intensity and capital requirements through centralized water and
waste-water processing; and,
89
Figure 5.1: Visualization of an EIN as described by Lambert and Boons (modified) [1]
• reduced raw material use intensity though the multi-facility exchange of co-products,
by-products or residual materials.
These are key factors in all eco-industrial park development and are documented as
such by a number of authors [1, 10, 110, 8, 7]. A visualization of these concepts is shown
in Figure 5.1 as presented by Lambert and Boons [1].
Life cycle assessment (LCA) is a technique typically used as part of an environmental
management system for analyzing the performance of a process based on the emissions
attributable to that process. Generally, a study conducted at a facility is compared to a
baseline of emissions established for a generic plant in the same industry. In this work,
LCA is used as part of the objective function in a mixed integer linear program (MILP)
simulation of an EIN to optimize it for reduced emissions. Another portion of the objective
function is attributed to financial gain from operating in an EIN compared to each plant
operating as a stand-alone facility. GAMS is used in this work to construct and solve the
90
MILP simulation to determine plant capacity considering the dual objective of reducing
emissions and maintaining or increasing profitability.
In recent years, several groups have attempted to apply LCA concepts to EIN arrange-
ments. A study on the Finnish forest industry revealed a potential benefit of 5-20% in most
impact categories considered by the LCA, and the authors state that “LCA seems a very
useful, albeit labor-intensive, tool for this kind of assessment. It can also help in detecting
those flows whose utilization could provide the greatest environmental benefits.” [111]. In
another study conducted by Mattila et al. in 2012, the authors address the methodology of
LCA applied to industrial symbioses, i.e., EINs [112]. The conclusion of the study is that,
to date, LCA has been applied in very few cases and also that “Expansion of current EIPs
and implementation of new ones may result in changes in the economic structure. This
change has not yet been analyzed in the IS [industrial symbiosis] literature, even though
LCA provides tools for such analysis.” [112]. To be clear, the work herein does not attempt
to analyze a change in economic structure or overall product outputs, but to consider that
reducing emissions and increased profits from industrial plants is a desirable outcome.
The eco-park considered in this work is shown in Figure 5.2 and is representative of
the material flow from each facility in the eco-park. The figure is not a traditional repre-
sentation of a network within an optimization context and instead represents the flow of
materials within the eco-park.
Previous work has been completed on a similar arrangement of processes and can be
found in a previous paper by Kantor et al. [69] which focused on a basic, preliminary de-
velopment of an optimization model in GAMS to optimize the network with a requirement
of hydrogen production to meet the demand of 1000 hydrogen fuel cell vehicles operating
91
Figure 5.2: EIN considered in this work [69]
within Ontario, Canada. The work presented in this chapter expands on the previous
model, introducing a more complex modeling technique of mixed-integer elements for ad-
ditional realistic consideration of production. This work uses the same baseline hydrogen
production as in the previous chapter but also furthers the analysis with the consideration
of the benefits of industrial integration using three different feedstocks for energy and reac-
tion components. The three fuels considered here are biomass, coal and natural gas. Each
fuel is considered separately, meaning that co-gasification of a mixture of biomass and coal
is not considered at this stage.
The decision variables for the optimization are:
• existence of each facility;
92
• existence of a connection between one facility and another;
• capacity of each facility; and,
• the division or ‘split’ of products from one facilities to others.
Also in this work, life-cycle assessment databases such as ecoInvent and the US LCA
database are used as a primary source for life-cycle data to improve upon the reliability
of the input data. These databases prescribe specific methodologies for obtaining and
publishing data and thus they are assumed to be more consistent over the broad range
of products considered. Additionally, each dataset is vetted for quality in the areas of
reliability, completeness, temporal correlation and geographical correlation as described by
Weidema and Wesnæs [113].
This work represents a growing part of the field of analyzing eco-parks and quan-
tifying their benefits. Several authors have expressed that an EIN has many benefits
[9, 16, 110, 8, 35] but empirical analysis of these benefits remains relatively unexplored
[112, 111]. Karlsson and Wolf utilize an optimization model to explore the benefits from
integrating a system comprised of a sawmill, pulp mill, district heating and biofuel upgrad-
ing [9]. Their method compares the baseline case of no integration with several cases of
integration between the different parts of the network. Additionally, the authors use the
terms industrial symbiosis and polygeneration synonymously with industrial integration,
as is common practice in the field of industrial ecology.
5.1.1 Eco-park concepts
The concept of an eco-park is described in several publications [9, 8, 10, 7] and is described
again briefly here to illustrate the idea. Eco-parks are a method of industrial cooperation
93
in which products, co-products and large centralized utilities can be operated to maximize
the efficiency of producing a wide array of outputs within a fixed geographical area [1].
Major improvements by operating in an EIN structure can be found in:
• large plants for water/waste treatment, heat exchange and electricity production;
• exchange of products and co-products between facilities [114, 2];
• waste reduction through facility intergration [114, 2]; and,
• increased optimized operation for higher profit or environmental performance [1].
It is the goal of this research to quantitatively exhibit the benefits of EINs in both
the economic realm as well as for reducing overall environmental impact of manufacturing
chemical products.
This research contributes to society by developing a method for evaluating industrial
relationships and by assisting in the planning of new facilities. This will benefit citizens as
the optimization will take environmental factors into account and will attempt to minimize
the overall waste and air emissions from facilities that could otherwise affect living condi-
tions in areas surrounding these facilities. The impacts on air, water and land can all be
considered and the importance of the environment is taken into account in addition to the
economic performance of industrial processes. If construction of a new chemical facility, or
more specifically a collection of facilities, can be made more economically feasible by its
integration with other processes in the region, construction and factory workers would also
be required to build and operate these facilities. This would contribute to the economic
stability in the region in addition to causing a decline in unemployment rates.
While environmental metrics are becoming increasingly important to business leaders,
companies are still responsible to their shareholders to show solid and sustainable economic
94
performance. The use of a dual-objective function allows for profitability to also be consid-
ered in the optimization and this will lead to a solution that proves to be environmentally
responsible as well as being economically feasible. This research is intended to reinvigorate
discussions in the industrial sector regarding issues such as sustainability, process symbiosis
and collaborative efforts. The EIN concept is synonymous with polygeneration, industrial
symbiosis and the like. All of these terms are based upon the concept of a diverse group of
industrial producers cooperating to achieve a common goal of cost-savings and/or reduced
environmental impact. Research has also shown that developing these eco-parks can very
much be a driver for innovation and development of new technologies [37, 38]. The Dow
Jones Sustainability Index (DJSI) is an indicator used to manage investment funds based
on sustainability metrics. The funds developed using this index showed solid growth since
the inception of the program and is an indicator that sustainability within an organization
can have significant impacts on profitability. Though the index also focuses on business
practices and management styles, the overarching reality is that sustainability within a
company yields financial performance results [39, 40]. The DJSI was considered as a po-
tential metric system for measuring the economic and environmental sustainability for the
eco-park network scenarios but was ultimately rejected due to qualitative parameters that
were not considered as part of the optimization model. The WAR algorithm, proposed by
Cabezas [99], was predominantly used in the construction of the metrics for optimization.
The premise of this system is utilized by researchers and government in order to assess
the amount of waste reduced from a process or process alternative [98, 97]. This approach
was used in combination with life-cycle assessment to construct the metric indices in the
objective function for reducing the overall waste reaching final disposition within the air,
water and land.
The approach for this optimization is to calculate the life-cycle impact of each product
or process in the proposed network and to compare the eco-park scenario with a plant of
95
comparable size operating as an independent facility. The eco-park concepts rely on col-
laboration from progressive facility managers in order to implement a symbiotic strategy
for responsible and sustainable chemical processing.
5.1.2 Life-cycle Assessment
Life cycle assessment can be used as part of an environmental management strategy to
assess and manage the life cycle inventory of emissions from a product or process. In this
work, LCA efforts are used in the construction of the objective function and for assessing
the reductions in environmental waste from operating in an integrated scenario. The life-
cycle inventory is taken from LCA studies and from SimaPro software databases. Life-cycle
impact assessment methods are used to relate emissions to potential impacts on people and
the environment as is described in the ISO 14040 series of standards [31].
SimaPro databases are region-specific but can be used as an estimate for emissions
from manufacturing a variety of products. Data quality considerations are monitored
within SimaPro according to the framework set out by Weidema and Wesnæs [113] and
the goal for this chapter was to obtain reliable, complete and temporally relevant data sets.
Although the life cycle inventories can respond to regional electricity generation and other
geographical considerations, less weight was applied to obtaining data specific to Canada.
For the figures below, cut-off values between one and six percent of total emissions are
applied so that only the major contributors to emissions are shown. If the figures were
not truncated as such, they would be unreadable as they would contain several thousand
elements contributing to the final emissions burden for the final product. As the cut-off
values are very small, the omitted elements are minor overall contributors to the life-cycle
emissions burden.
96
Figure 5.3 shows the life-cycle greenhouse gas (GHG) contributions from various sources
required to produce 1 kg of ammonia. This is a representation of the GHG emission contri-
bution from different aspects of ammonia production using steam-methane reforming which
is a standard practice for producing ammonia. The cut-off applied in this instance for rep-
resenting the major emission contributors is 1%. In this typical production of ammonia, it
can be observed that natural gas production and use, in several stages, is responsible for the
majority of GHG gas emissions. It must be noted, however, that the emission landscape
changes depending on the emission being considered. For example, when considering the
emission of sulphur oxides, the emissions from the natural gas streams and fuel oil streams
are similar as shown in Figure 5.4 with an applied cut-off value of 3% in order to properly
view the network of life-cycle contributions. This is logically sound as there is typically
more sulphur contained in, and released from, crude oil when compared with natural gas
[115].
The complexities involved with assessing the emission burden from each product /pro-
cess requires an objective function that will take each emission into account and its relative
importance to society. The scope of this work covers the emissions of GHGs, pre-cursors
to photochemical smog, oxides of sulphur and solid waste.
Figure 5.3 and Figure 5.4 are examples of the life-cycle assessment where the CO2 and
SOx emissions associated with producing ammonia in a traditional facility which is not
part of an EIN are evaluated. The purpose of these examples is to show the areas in which
the production of ammonia can be improved in order to decrease life-cycle emissions. The
thickness of arrows exhibits the contribution from one area of the life cycle production pro-
cess to the overall emissions burden associated with a product, in this case, ammonia. For
97
Figure 5.3: Life-cycle GHG contributions for producing Ammonia via a traditional (non-
EIN) process from EcoInvent Database of SimaPro software
98
Figure 5.4: Life-cycle SOx contributions for producing ammonia via a traditional (non-
EIN) process from EcoInvent Database of SimaPro software
example, it can be seen in Figure 5.3 that the greatest contribution to GHG emissions in
the ammonia production process is the steam reforming of natural gas. Minor contributors
in this case are generation of electricity and fuel oil for heating and transportation. Figure
5.4 shows the life-cycle emissions of SOx and shows a much more even split of the pro-
cesses contributing to SOx emissions stemming from ammonia production; furthermore,
it exhibits that the requirements of fuel oil, natural gas and nickel are the primary areas
contributing to these emissions. It is important to understand life-cycle concepts when
attempting to integrate processes into an eco-park network as the major contributors to
emissions of a particular type are the best candidates for improving the environmental
performance of that process.
99
For both instances, the network layer previous to the finished product represents the
sum of the process production flow charts to create the product or process used by the final
manufacturing. Discrepancies between the sum of the penultimate layer and the final life
cycle emission contribution are attributed to the processing within the plant, defined as
gate-to-gate emissions for the facility. Figure 5.3 shows high gate-to-gate GHG emissions
and thus represents a large opportunity for reducing the life-cycle GHG emissions within
the plant. Figure 5.4; however, shows most of the life-cycle SOx emissions are accounted
for prior to entering the manufacturing facility. The opportunity for reducing the life-cycle
SOx emissions is thus bound with the process feedstock.
Researchers have only begun investigating the possibilities of evaluating EINs using
LCA concepts, this work not only shows that this is a valuable undertaking but also pro-
ceeds to utilize optimization in order to assess the best way of constructing these facilities
based on the LCA concepts.
5.2 Manufacturing Facilities
The model is formulated as an MILP with chemical reactions, conversions, product removal
and recycling. Supply and demand are modeled as in a classical transportation problem
but varies significantly due to reaction and/or separation at each facility or ‘node’. The
mass and energy balances must be written for each node to account for the flows of en-
ergy and material through the network. Following this, the technical constraints must be
quantified in a mathematical format in order to implement them within the model.
The nodes of this network were chosen in order to process the streams of a fuel source
100
into products. Gasification processes (G) can be constructed to accept input of coal or
biomass and produce syngas of varying H2 : CO, and CO : CO2 ratios. The alternative
fuel source considered is natural gas, with the syngas mixture achieved by steam-methane
reforming. Some of the syngas can be used directly in a combined heat and power plant
(CHP) to produce heating and electricity for the network processes.
For the remaining nodes in the network, the hydrogen in the syngas must be separated
from the other gases in order to utilize them in further processing stages. The separation
of these gases is completed by the pressure-swing adsorption process (PSA). To ensure
that the network operates with the least possible emissions, a carbon capture process (CC)
using a recirculating ammonia loop is added to the network to capture CO2 emissions
which would otherwise be emitted to the atmosphere. A CO2 sequestration process (SQ)
is added as a possibility for the network to reduce its emissions to the environment. This
node relies on injection of CO2 into a deep saline aquifer as is the most feasible form of
geological storage in Ontario as explored by Shafeen et al. [116]. It is estimated in the
same work that the storage capacity via geological sequestration is approximately sufficient
for 730 million tonnes of CO2. For the scale of processing considered in this work, the stor-
age capacity is much larger than the amount produced; however, the constraint is placed
within the model in the event that the scale of facilities considered is increased significantly.
Ammonia production (AM) in the network is sufficient to supply the carbon-capture
process with make-up ammonia while also producing excess as a market product which is
used as a chemical building block for other processes or as an agricultural fertilizer. Urea
processing is naturally synergistic to the ammonia process as urea requires two molecules
of ammonia and one molecule of CO2. The ammonia can be produced at the proper con-
ditions for urea production, negating additional processing considerations for conventional
ammonia destined for use in manufacturing urea. The remainder of the ammonia is sent
101
to market.
A urea node is included in the network as it is efficient to produce ammonia in close
proximity to ammonia manufacture [117]. Urea is also a chemical fertilizer which can be
used to temporarily sequester CO2 as part of a solid fertilizer. This CO2 is later released
to the atmosphere when urea is applied as a fertilizer but the emission of CO2 in close
proximity to the vegetation can increase probability of its utilization for respiration. Urea
manufacturing (U) in this network is entirely for export to external markets. A summary
of the network nodes is shown in Table 5.1. Table 5.2 shows a summary of the inputs and
outputs for each facility.
Table 5.1: Legend of Nodes
Process Node Abbreviation
Gasification G
Combined Heat and Power CHP
Carbon Sequestration SQ
CO2 Capture CC
Pressure-swing Adsorption PSA
Ammonia Production AM
Urea Manufacture U
5.3 Objective Function
As mentioned previously, the complex nature of the life-cycle emission considerations must
be included in an objective function that will lead to the optimization of the network of
102
Table 5.2: Summary of inputs and outputs
Process Inputs Outputs
Gasification
Biomass/Coal/Natural gas CO
Heat CO2
Electricity N2
Air H2
H2O
Combined Heat and Power
CO Heat
CO2 Electricity
N2 N2
H2 H2O
Heat CO2
Electricity
Carbon Sequestration
CO2 (purified stream)
CO2 (purified stream)Heat
Electricity
CO2 Capture
CO2 CO2 (purified stream)
N2 N2
H2O H2O
Heat
Electricity
Pressure-swing Adsorption
CO H2 (purified stream)
CO2 CO
N2 CO2
H2 N2
Heat H2
Electricity
Ammonia Production
N2
NH3
H2
Heat
Electricity
Urea Manufacture
NH3
(NH3)2CO2
CO2
Heat
Electricity
103
plants yielding the most beneficial emission reductions. A purely environmental objective
function is impractical for two reasons; primarily, a purely environmental objective is
unlikely to be undertaken by industrial interests as the manufacturers must show positive
financial results. Additionally, it is impractical to consider a purely environmental objective
as the optimization algorithm would naturally decrease the plant sizes to the least allowable
value. This assessment does not include the emissions offset associated with reforestation.
For these reasons, an economic term is included to offset the environmental term to present
a balanced, applicable, practical approach to optimizing the network. The bi-objective
optimization is structured as described by Kim and Weck [118] and the formulation for the
program is shown below.
For an optimization program to minimize Z subject to a design vector x and a vector
of fixed parameters, c, the objective function can be written as equation 5.1.
minZ(x, c) (5.1)
The entire problem can be structured as follows:
minZ(x, c)
s.t. g(x, c) ≤ 0
h(x, c) = 0
xLi ≤ xi ≤ xUi (i = 1, . . . , n)
Z = [Z1(x) . . . Zq(x)]T (5.2)
x = [x1 . . . xi . . . xn]T
g = [g1(x) . . . gm1(x)]T
h = [h1(x) . . . hm2(x)]T
Where:
104
• g is the vector of inequality constraints;
• h is the vector of equality constraints;
• xLi is the lower bound for xi;
• xUi is the upper bound for xi;
• q is the number of objectives;
• x is the vector of decision variables;
• c is the vector of fixes parameters;
• m1 is the number of inequalities; and,
• m2 is the number of equalities;
and with these definitions, 5.2 can be reduced to a scalar problem of the form shown
in 5.3
min Z =
q∑i=1
λiFiZi (5.3)
Where λi and Fi are the weighting and scaling factors for each Zi. Z is considered to be
the aggregated objective value, being a summation of each weighted element Zi as shown
by Kim and Weck [118]. Generally, the sum of the weighting/scaling ratios is equal to
unity. Kim and Weck explored this structure in the case of a bi-objective function which is
applicable in this work, as the two objectives being considered are the economic objective
and the environmental objective which is written in terms of reduced emissions.
Splitting the objective function into the economic portion and an environmental por-
tion yields Z1 = Zeconomic and Z2 = Zemissions, respectively. Though this form would then
105
appear as a bi-objective function, Zemissions is constructed inherently as a multi-objective
function to include the emissions of multiple environmental contaminants. Zemissions is
constructed as a summation of the reduced emissions between an independently-operated
facility when compared to the integrated facility such that:
Zemissions =ne∑e
λeFeZe (5.4)
Where:
• e is representative of a considered emission;
• ne is the number of emissions considered;
• λe and Fe are the weighting and scaling factors as mentioned previously; and,
• Ze is the emission differential between the integrated and stand-alone facilities. The
difference is defined in Equation 5.5 below:
Ze = Ie − Se (5.5)
Where:
• Ie is the emissions of e from an integrated facility;
• Se is the emissions of e from a standalone facility.
Constructing the economic portion of the objective does not require further manipulation,
as it is a difference between the net present value (NPV) of the integrated and independent
plants assessed at a set discount rate, rd, and plant lifetime and is constructed as shown
106
in Equation 5.6.
Zeconomic = ICC − SCC +L∑t=1
np∑p
Rp − [IOC − SOC ]p(1 + rd)t
(5.6)
Where:
• p is a manufacturing facility;
• np is the number of manufacturing facilities;
• Rp is the return from sale of products from plant p ($);
• ICC and SCC represent the integrated and standalone capital costs, respectively($);
• IOC and SOC represent the integrated and standalone operating costs, respectively($);
• t represents the year; and,
• L represents the lifetime of plant p.
The numerator in equation 5.6 does not require the subscript t, as it is assumed that
production is maintained at the same level for the lifetime of the plant. For this analysis,
the network lifetime, L, is considered to be 30 years during which time, the discount rate,
rd, is also fixed. The full bi-objective function, based on these equations can then be seen
in equation 5.7.
min Z =λemissionsFemissions
Zemissions +
[1− λemissions
Femissions
]Zeconomic (5.7)
107
5.4 Modeling
The modeling is explained in detail in a previous chapter. The explanation here is pre-
sented to summarize the main differences in the model which are added in order to create
a more realistic scenario with mixed-integer programming. A schematic of the network
under consideration was shown previously as Figure 5.2.
The sets used in this formulation are as follows:
• k is a set of the transportation technologies available;
• m is a set of plant sizes;
• v is a set of material vectors;
• e is a set of emissions;
• p is a set of facilities; and,
• p2 is an alias of p.
The mass balances between units are formulated as inequalities as described by the vector
g in Equation 5.2, whereas the mass balances within a unit are formulated as equalities
as described by vector h in Equation 5.2. The format of some inequalities in the following
equations is altered from the general format for the sake of clarity. The generalized form
of the mass balances between units are described by Equation 5.9, while the mass balances
within a unit are described by Equation 5.8.
outv,p =∑p
inv,p − rxv,p + genv,p ∀v, p (5.8)
108
Where outv,p is the amount of material v output from plant p and inv,p is the input of v
to plant p. This equation is a mass balance based on equilibrium conditions within a unit
such that the amount output from a given plant p is equal to the inputs from all other
plants p2 with a decrease due to consumption at a rate of rxv,p or an increase from genera-
tion encompassed by genv,p. rxv,p and genv,p are functions of conversion and stoichiometry.
outv,p ≥∑p26=p
inv,p2 ∀v, p (5.9)
This mass balance between units forces the production of material v from plant p to
exceed the requirements of input for all other plants, p2, that accept material v from p. At
this point, the binary variable x is defined for selection of plant sizing.
xp,m =
1 if plant p of size m exists
0 Otherwise(5.10)
To force the selection of only one plant, Equation 5.11 is included:∑m
xp,m ≤ 1 ∀p (5.11)
Additionally, the flow of material v from plant p must be less than the capacity ex-
pressed by xp,m. This constraint is applied by utilizing Equation 5.12.
outv,p ≤MUp,mxp,m ∀v, p,m (5.12)
Where MUp,m is the upper limit of from plant p of size m. Each value of MU
p,m is defined
in a table within the optimization program.
109
At which point ICC and SACC from Equation 5.6 are calculated for plant p of size m
from existing plant data for the stand-alone and integrated facilities. The economy of scale
plant construction theory with an exponential scaling factor of 0.6 is used to calculate the
costs for sizes lying between defined points. The costs are amortized and included in the
economic portion of the objective function. The existence of a transportation connection
between two plants is represented by variable y as described by equation 5.13.
yp,p2,k,v =
1 if material v is transported by method k between p and p2
0 Otherwise
(5.13)
Several integer equations must be applied at this time. The first, Equation 5.14, pre-
vents the existence of transportation connections to plant p if plant p does not exist in any
size (i.e.,∑m
xp,m = 0).
∑m
xp,m ≥ yp,p2,k,v ∀p, p2, k, v (5.14)
Equation 5.15 yields the result that only one method of transportation should be chosen
to transport material v between plants p and p2.
∑k
yp,p2,k,v ≤ 1 ∀p, p2, v (5.15)
The transportation cost can thus be assessed by applying the base-plus-throughput
method. This is shown here in Equation 5.16.
110
TransportationCostk =∑p
∑p2
∑v
yp,p2,k,vBaseCostk+∑v
∑p
outv,pThroughputCostk ∀k
(5.16)
The transportation cost, in turn, is also included in the economic portion of the ob-
jective function. More details on the interactions of each unit and the associated reagents
can be found in Chapter 4.
The structure of the environmental portion of the program is constructed in a similar
manner to the economic portion shown above, yet is increasingly complex as each emission
type (set e, noted above) requires its own correlations. These correlations are parallel to
the economic portion of the objective function, yet they yield the differences in emissions
produced from a standalone facility and an integrated facility instead of the economic dif-
ferences.
5.5 Results
The EIN considered here is to support production of hydrogen for export, in anticipation
of the potential ‘hydrogen economy’ where there is some demand to refuel fuel cell vehicles
[41]. The target for hydrogen production in the EIN is enough to fuel 1000 vehicles within
Ontario, Canada. As presented in a previous chapter, the results of the model show that
to produce the requirement for 1000 vehicles fuelled by hydrogen, the export of products
from facilities will vary based on the weighting factor of life cycle emissions in the objective
function [69]. By varying the life-cycle emissions weighting in the range of 0, representing
no impact of emissions on the objective function, to 1, representing no impact of profitabil-
ity on the objective function yields the result shown by Figure 5.5. The x-axis in Figure
111
Figure 5.5: Normalized plant exports as a function of life-cycle emissions weighting
5.5 refers to the ratio λemissions
Femissionsas mentioned in equation 5.7 and refers to the weighting of
the life-cycle emissions relative to the economic weighting. As it is a bi-objective function,
the two sides of the figure are noted as being the side of high environmental bias or high
economic bias.
Normalized plant export, used on the y-axis of Figure 5.5- 5.7 is representative of the
output from each facility in the EIP. The values are normalized to the maximum value
seen over the trials for the purposes of observing changes in production from each facility.
As the level of production varies largely between plants, a figure of the normalized values
is best for observing variations in the level for each facility.
As suggested by previous work, the possibilities of alternative fuel sources should be
112
considered [69]. For this work, coal and natural gas have been considered as possible al-
ternatives to biomass as the primary fuel source for gasification. Modifying the model
to accept a variety of fuel sources as inputs to the eco-park shows any potential benefits
which may arise from utilizing a traditional fossil fuel as a gasification or steam-methane
reforming (SMR) fuel, providing that the network still provide an adequate level of hy-
drogen for 1000 fuel cell vehicles. Figure 5.6 and Figure 5.7 show the same analysis but
using coal or natural gas as fuels. Figure 5.6 shows that importing electricity is required
for emission weighting between 0 and 0.8 in order to produce the required products. This
is likely due to a balancing of emissions from using coal generation and the grid mix in
order to supply the necessary electricity to the network. High volumes of coal are required
at a high value of emissions weighting as coal can be utilized within the network in a
more emission-efficient manner than the electricity production that would be imported
into the network in the low-emission-weighting scenarios. The highest production for most
industrial chemicals is at an emissions weighting of 0.9 at which point, emissions have a
large impact but economics also plays a role. At higher emissions weighting, the economic
portion of the objective is no longer used and thus the production drops off significantly in
order to reduce the amount of emissions while maintaining the level of hydrogen production
for the fuel cell vehicles as stipulated previously.
Figure 5.7 represents the same information as in the previous two figures but utilizing
natural gas as the feedstock. There are three distinct regions, the first when the life-cycle
weighting is between 0 and 0.4, the second between 0.5-0.9 and the final region when the
life-cycle emissions are weighted as 1, representing a purely environmental objective. Sim-
ilar trends are seen as with the coal except that electricity import is only occurring in the
scenario in which life-cycle emissions weighting is set to 1. At this point, the optimization
again reduces the amount of chemical production as economics are no longer a factor. In the
intermediate to high range of emissions weighting, 0.5-0.9, large product volumes are man-
113
Figure 5.6: Normalized network plant export using coal as gasification feedstock
ufactured in the EIN as there is an economic incentive in addition to the reduced emissions.
Also, the analysis points to the processing of urea being most economically viable when
balancing emissions and economics. Urea is a value-added product and is easier to produce
using a natural gas feedstock than other potential sources. The CO2 that is bound in the
urea is also beneficial economically as the carbon sequestration facility would experience
reduced loading and thus would require lower capital and operating costs.
The weighting for life-cycle emissions and profitability were set to the baseline level of
0.5 for each factor. Figure 5.8 and Figure 5.9 show the results of the updated modeling in
terms of production amounts, profits and emissions for the three fuels.
114
Figure 5.7: Normalized network plant export using natural gas as feedstock
In terms of profitability, the model results shown in Figure 5.8 yields that biomass is
the least profitable fuel to use and that natural gas yields slightly higher profit than coal.
The reasoning for a slight but marginal difference using the various fuels is due to a dif-
fering network price structure considering SMR instead of gasification of coal or biomass.
The biomass yields a lower net profit due to a lower energy density and increased pro-
cessing complexities associated with utilizing biomass in the system. The total emissions,
surprisingly, are similar for all three fuels. Any potential emission reduction from utilizing
biomass as a feed is offset by the increased transportation emissions of the biomass and
the fact that reforestation has not been considered as it is not part of the EIN operations.
It is assumed that inclusion of reforestation to re-sequester carbon through the growth of
new biomass over the EIN lifetime would reduce the total emissions produced by utilizing
biomass as a fuel. The hydrogen, as mentioned, has a fixed level in order to produce the
115
Figure 5.8: Network product exports and profit for three different fuels
required fuel for 1000 vehicles and thus it is constant across the three fuels considered. The
production level of ammonia varied slightly for the three fuels, with the largest amount
being produced while utilizing biomass as gasification feed.
Figure 5.9 shows the heat, power and urea output for the three fuels. Immediately, it
can be seen that the net power export for using biomass is a negative value, indicating
that power import is required. Both fossil fuels show a net export of power, although the
level is very low. Clearly the low energy content of the biomass requires electrical input to
the EIN. Heat output is also dramatically lower for biomass than for coal or natural gas,
although this was expected due to the lower energy content in the fuel. The only fuel which
showed a meaningful production of urea is natural gas. Due to the inclusion of life-cycle
emissions as part of the objective function, ammonia is typically produced in higher quan-
116
Figure 5.9: Network exports of power, heat and urea for three different fuels
tity than urea as the additional processing of ammonia to produce urea generally does not
produce a favourable cost/benefit transaction in terms of the objective function. With the
case of natural gas, the cost of erecting the urea facility is only slightly overcome by the
profitability associated with the product. For the other fuels, this barrier is not exceeded
and thus urea is not produced in any significant quantity.
Calculating the mass-specific contributions of ammonia production between the tradi-
tional process shown in Figure 5.3 for GHG emissions and Figure 5.4 for SOx emissions
yields the comparative tables shown in Table 5.3 and Table 5.4. While the differences in
Table 5.3 seem minute, Table 5.4 is shown for comparison at a level representing a very
small, considering typical industrial operations, production of ammonia. The reduction in
significant figures for Table 5.4 belies some of the differences due to rounding; whereas,
117
the benefits in terms of GHG emissions are very clear. Some issues, such as allocation
of emissions to the appropriate facility, are sometimes difficult to address and thus it is
generally a better approach to look at the emissions from the network perspective instead
of each individual product. Regardless of the possible allocation issues, the EIN shows
large reductions in GHG emissions using any of the three fuels and mixed results for SOx,
smog precursors and solid waste. Part of this reduction is due to the incorporation of car-
bon capture and sequestration within the network but reductions in transportation-related
emissions and fugitive releases also contribute.
Table 5.3: Comparison of life cycle emissions per kg of ammonia based on three fuels
Basis: 1 kg Ammonia Traditional EIN with EIN with EIN with
Natural Gas Coal Biomass
GHG (kg CO2 equivalent) 1.91 1.1469 1.14125 1.00464
SOx (kg SO2 equivalent) 0.00333 0.00293 0.00405 0.00146
Smog Precursors (kg VOC
equivalent)
0.00309 0.00329 0.00308 0.00437
Solid Waste (kg) 0.0033 0.00326 0.00386 0.00396
Total 1.91972 1.15637 1.15224 1.01443
5.6 Summary
This chapter focused on the expansion of the model from the previous chapter to include
integer and binary variables for options of plant size/existence and connections between
the facilities. In addition, the gasification node was implemented in the model and three
fuel sources were considered. Natural gas, coal and biomass showed similar results for
118
Table 5.4: Life-cycle emissions of ammonia based on three fuels at a reference level of 730
tonnes
Basis: 730000 kg Ammonia Traditional EIN with EIN with EIN with
Natural Gas Coal Biomass
GHG (kg CO2 equivalent) 1400000 840000 840000 740000
SOx (kg SO2 equivalent) 2400 2400 3000 1100
Smog Precursors (kg VOC
equivalent)
2300 2400 2300 3200
Solid Waste (kg) 2400 2400 2800 2900
Total 1410000 850000 840000 740000
many of the plant exports although the lower heat content from biomass coupled with
additional impurities led to reduced profitability and less generation of heat and electricity
when compared to the other two fuel options.
The sensitivity analysis on the weighting of each portion of the objective function was
carried out for each fuel and result in four distinct optimal scenarios with varying levels of
production from each of the facilities.
Compared to the previous chapter, the overall economic and environmental benefits are
slightly lower which is attributable to the increasing constraints imposed in the model such
as discrete plant sizes and inclusion of transportation costs. The MILP approach presented
in this chapter allows for increasingly detailed modeling of the EIN than the LP presented
in Chapter 4 and is a more realistic representation of facility construction.
119
Chapter 6
Generalized MINLP Modeling of
Eco-Industrial Networks
6.1 Introduction
Symbiotic relationships between chemical network producers have been demonstrated ef-
fectively in both academic and industrial domains. The integration of material and energy
exchange between producers typically offers economic and environmental advantages over
stand-alone facilities. However, the high level of interactions between participants in an
eco-industrial network (EIN) requires the application of a system integration method for
the design of a common optimal infrastructure. One of the crucial factors for the success
of any EIN is that the sum of the benefits achieved working together must be greater than
working as many stand-alone facilities [119]. For example, the industrial symbiotic network
of Kalundborg, created in the 1970s, is considered a prototype of an EIN. This network
consists of an oil refinery, a pharmaceutical company, an electric power plant, a gypsum
plate factory, a cement factory, a fish nursery and city heating. These facilities use surplus
energy and waste materials from each other, obtaining annual savings of more than $12
120
million [120]. Currently, the government of South Korea is implementing a three-phase,
15-year EIN initiative to promote the balance between the economy, society and environ-
ment [121, 8, 122]. The first phase of the project (2005-2009) aimed at converting five
existing industrial complexes into EIN based on the optimization of energy consumption,
raw materials, and other resources. This is part of the South Korean “low-carbon green
growth” strategy, which has become the core paradigm of development since 2008 [123].
Accordingly, the focus of these networks has been analyzed qualitatively by a number
of authors and quantitative examples have now been presented in a number of instances.
Lovelady et al. [119] developed an optimization model for the design and integration
of EINs. The model focuses on the management of water among multiple processes in
a common industrial network. The recycle, reuse, and separation of waste-water using
interception devices are considered as management strategies. The model is based on a
source-interception-sink structure to determine the best potential configuration. Fernandez
et al. [124] proposed a model to determine the optimal location for a sustainable EIN. The
model is based on the analytical hierarchy process (AHP) that applies multicriteria eval-
uation to analyze different suitable locations. The model consists of three different levels:
selection of the geographic area, evaluation and selection of suitable areas, and evaluation
of specific zones. The evaluation is performed according to the importance assigned to the
goal variables (i.e., social, economic, environment, planning, and infrastructure). There-
fore, weights are applied to the different factors considered in the model, which leads to
the identification of the suitable locations.
Sendra et al. [125] adapted a material flow analysis tool to consider material flow into
and out of an EIN. The model determines the amount of material and energy that are
used in the network to plan the development of an eco-park. The model also takes into
account environmental metrics for the analysis. Zhao et al. [126] used system dynamics
121
and grey clusters approaches to redesign an EIN in China. Four different scenarios were
considered in the analysis: base case, economic development as key priority, impact on
the environment, and economic impact (subject to science, technology, and environmental
improvements). Spriggs et al. [127] described the challenges involved in the development
of eco-parks. The authors classified these challenges into two groups: technical/economic
and organizational/commercial/political. The technical/economic challenge means that
the success of an eco-park greatly depends on the feasible exchange of materials and energy
among the companies involved in the network. The organizational/commercial/political
challenge means that deep cooperation/integration amongst the companies in the EIN are
needed to set common rules that lead to the optimal flow of information in the network.
Additionally, Chertow [4] proposed taxonomy for EINs based on the type of material ex-
change. He studied 18 potential networks and suggested the following classification: waste
exchange within a facility/organization, among facilities co-located in a defined eco-park,
among facilities at some distance (e.g., Kalundborg) and among facilities structured across
a broader region. Another approach presented by Baldwin et al. [128] who studied various
EIN models (Kalundborg, Styria and Massachusetts) based on an evolutionary framework.
This work represents a novel approach to the construction of such networks that includes
optimization of material and energy streams and uses a purely quantitative methodology
for evaluating these eco-park scenarios. This work expands on the eco-park principles
to include additional mathematical complexity in the model formulation as well as ana-
lytical insight into the network’s design. Kantor et al. [69] developed a linear program
(LP) that models an EIN to demonstrate its economic and environmental feasibility and
the approach used to solve this type of problem. The model considered a dual-objective
function to account for economic and environmental factors existing in processing facilities.
In this chapter, the base models presented in the previous chapters have been refor-
122
mulated to add additional complexity and express the life-cycle effects of a transporta-
tion network for materials and energy between the production nodes. The extension of
this model has required a shift into the realm of mixed integer non-linear programming
(MINLP). This type of model allows for increased flexibility in constraint construction and
improved accuracy of the model. Shifting to MINLP methods also has consequences, which
include: increased computational time, less well-defined algorithms for providing solutions
and concerns regarding the globality of a solution, once found.
Life cycle concepts are particularly useful in determining environmental benefits of
alternatives as the methodology is capable of capturing emissions produced from resource
extraction to final disposition [129, 130]. Accordingly, for further development in terms
of transportation analysis and improvements, the model was reformulated as an MINLP
model to design this EIN. Additionally, the objective function of the model has been
reformulated in such a way to prevent either of its two terms from overshadowing the
other.
6.2 Multi-Objective Optimization Model
This section presents the main features of the multi-objective optimization model pro-
posed in this work to determine the annual production costs (accounting for environmental
metrics) associated with the EIN. The model has been constructed using GAMS, as dis-
cussed previously, as it has been used extensively for multi-objective optimization as well
as MINLP modeling.
123
6.2.1 Problem Statement
The multi-objective model proposed in this work aims to minimize the life-cycle emissions
while attaining economic profits in an EIN design. Figure 6.1 shows the general layout
of the present network design model. The model consists of nodes defined by different
industrial processes designed to operate as a network. These nodes exchange material
and/or energy, thus minimizing waste (by reutilizing process products and co-products from
other EIN facilities) and maximizing profit (by generating higher added-value products)
under operational constraints. As a result, the integrated relationship between nodes
has been considered in the present mathematical formulation. The model’s key inputs
are: maximum output capacities of the production nodes, fuel composition and distances
between the nodes. To meet the expected production levels of energy carriers and chemical
products, the present EIN simultaneously minimizes the environmental impacts (in terms of
life cycle emissions) and production costs by selecting the most suitable type of feedstock
fuel, nodes in the production network and transportation modes subject to operational
constraints. The model’s key outputs are (see Figure 6.1): annualized network production
costs, type and amount of products, plants and systems with corresponding capacities, and
overall economic and environmental performance relative to non-integrated facilities.
The present EIN model includes the selection of continuous, binary and integer vari-
ables. In addition to these variable types, there are non-linearities present in the model
from various constraints and calculations; thus, the resulting mathematical model is formu-
lated as an MINLP optimization problem. Each of the processes (i.e., production nodes)
considered in the present model (see Figure 6.1) are described in detail in the present sec-
tion.
A formulation of this network was initially constructed as an LP model which relies
on continuous variables without complex interactions between variables. This type of
124
Figure 6.1: Eco-industrial network flow diagram
model is the most rigorously studied and most easily solved type of optimization. For this
reason, large bodies of research have been published on this type of model and for lineariz-
ing complex constraints to take advantage of well-known algorithms for solving LP models.
Typically, LP models are incapable of capturing the full extent of a complex opti-
mization and thus the MILP type of optimization is used. This type of model allows
for binary and integer constraints to be used in the model which permits discrete deci-
sions and discontinuous variables to be used in the model. This type of model is used for
more extensive optimization problems as it allows the user to construct a more realistic
model of the system. As with LP models, MILP models often rely on linearization tech-
niques applied to more complicated constraints in order to maintain the MILP structure
125
to comply with the solution methods. LP and MILP methods are well-studied and require
relatively little computational time and power when compared to an MINLP formulation.
The solvers which are included in many standard optimization packages have been refined
to solve LP and MILP problems efficiently and effectively. MINLP modeling, however,
introduces non-linear constraints to an MILP model and is typically considered when the
constraints can no longer be linearized. The consequences of delving into this realm are
less well-defined solution algorithms and a large increase in computational power and time.
In this work, the nature of the constraints and the objective function would not allow for
a definite linearization and thus the model was necessarily constructed as an MINLP. The
benefits of utilizing this framework are that the model can be presented in a more realistic
and comprehensive fashion without compromising on the specificity of constraints or the
objective function.
As outlined in the previous chapters, this network of facilities was selected because
of natural integration that is experienced between these facilities. The network contains
products and co-products which act synergistically with other facilities in the network;
thus, locating such industries and collaborating toward an end goal of increased profits
and reduced emissions can be realized.
The base scenario for the optimization models is that each facility operates indepen-
dently of each other, i.e., no integration between the nodes. The integrated case is always
compared to this base case in order to derive a benefit of reduced emissions or improved
economics compared to the case without integration.
126
6.2.2 Problem Representation
To simplify the terminology, key indices have been introduced into the problems formula-
tion to represent the main properties of the model. A source-sink structural representation
is used to embed potential configurations of interest. Accordingly, the indices i and j have
been included to represent the source and sink nodes, respectively. The sets i and j are
defined as follows:
i ∈ I = g, chp, cc, psa, sq, ap, up, mk, wt (6.1)
j ∈ J = g, chp, cc, psa, sq, ap, up, mk, wt (6.2)
where g represents gasification processes, chp combined heat and power plants, cc
carbon capture, psa pressure swing adsorption, sq sequestration, ap ammonia production,
up urea production, mk the market place, and wt water and waste treatment facilities.
Also, a constraint has been included in the model to avoid material back-feeding in the
nodes (i.e., i 6= j). The set of alternative feedstock hydrocarbon fuels, h, used in the
network is given as:
h ∈ H = C, B, NG (6.3)
where h denotes the set of feedstock fuels available in the eco-industrial network, C
represents coal, B biomass and NG natural gas. Furthermore, the index m is used to
denote the set of materials, which includes products and co-products involved in the eco-
industrial network. The set of materials, M, is defined as follows:
m ∈M = M1, ...,M11 (6.4)
127
where M1 represents carbon monoxide (CO), M2 carbon dioxide (CO2), M3 hydrogen
(H2), M4 nitrogen (N2), M5 ammonia (NH3), M6 urea, M7 water (H2O), M8 methane
(CH4), M9 ash, M10 sulfur (S), and M11 other contaminants. On the other hand, the
set of utilities, u, associated to the eco-industrial network are heat, process water and
electricity. This utility set is denoted as follows:
u ∈ U = U1, U2, U3 (6.5)
where U1 represents heat, U2 process water and U3 electricity. Additionally, the set of
transport modes, k, available to connect the nodes is given as:
k ∈ K = Rd,Rl, Pd (6.6)
where Rd represents roadway transportation, Rl rail and Pd pipeline of diameter d.
Six different pipeline sizes are considered in the formulation of the model, ranging from
a minimum of two inches (2”) to a maximum of twelve inches (12”) with a difference of
two inches between each other (i.e., P2”, P4”, ..., P12”). The environmental emissions, e,
considered for the life-cycle emission assessment of this study are given by the following
set:
e ∈ E = GHG,NOx, SOx, SW (6.7)
where index e denotes the set of emissions, GHG represents the greenhouse gases mea-
sured in CO2 equivalent units, NOx are the nitrogen oxides, SOx are the sulfur oxides,
and SW represent the solid waste materials (landfill materials).
128
Figure 6.2: Gasification process
6.2.3 Model Formulation
As previously mentioned, the optimization model presented in this work consists of differ-
ent production nodes that can be linked to each other depending on material needs and
operational constraints. These nodes i and j are linked by different available transport
modes k depending on economic factors and the nature of the transported materials. The
EIN nodes considered in the model are described next.
6.2.3.1 Gasification node (g)
The model’s first stage consists of a gasification process, where the feedstock hydrocarbon
fuel, h, reacts at high temperatures generating a mixture of gaseous products. The gaseous
products are typically carbon monoxide (CO), carbon dioxide (CO2), water (H2O), hy-
drogen gas (H2) and nitrogen products (e.g., when air is used as the oxygen source of the
process). The diagram of this process is shown in Figure 6.2.
Hydrogen gas and carbon monoxide are the main outputs of the process and they are
available, as well as the rest of the products, for further use as chemical feedstocks in down-
129
stream nodes j [69] . The type of feedstock fuel and corresponding product composition
in the gasifier are calculated as follows:
GPm =fhFCh,mFM
MWh
(6.8)
fh =
1 if fuel h is selected
0 Otherwise(6.9)
where GPm represents the total amount of material m produced in the gasifier, FCh,m
is a parameter that represents the material composition of the gaseous product formed
in the gasification process as a function of the type of fuel h based on the literature
[131, 132, 133, 78, 134], FM is the mass flowrate of the fuel entering the gasifier, and
MWh is the molecular weight of the feedstock fuel. The selection of only one type of fuel
for the gasifier is constrained as follows:
∑h
fh = 1 (6.10)
where the index h represents the type of feedstock hydrocarbon fuel and fh is a binary
variable indicating the type of fuel h. The waste products (i.e., M9 −M11) obtained from
the gasification process are sent to waste/water treatment facilities (wt). These waste
products are given as follows:
GWm = GPm ∀m ∈ 9, 10, 11 (6.11)
where GWm represents the amount of waste materials generated in the gasifier (source
node) and sent to waste/water treatment facilities (sink node) for environmental treatment
and later disposal/commercialization. The rest of the materials generated in the gasifier
130
are co-products that can be used in the remaining production nodes for further processing,
these materials can be calculated as follows:
GOm,j = GSFjGPm ∀m ∈ 1, 2, 3, 4 , j ∈ cc, chp, psa (6.12)
where GOm,j represents the material m output from the gasifier to the production nodes
j, and GSFj is the gasifiers split factor to the sink nodes j. Furthermore, the gasification
process demands certain types of energy such as heat, water and electricity. These utility
demands (GDu) associated to the gasifier can be estimated as follows:
GDu = GRu
4∑m=1
GPm (6.13)
where GRu is a parameter that denotes the amount of utility u required in the gasifier
per unit of total gaseous product generated in the plant as defined by the requirements of
the facility operations as found in the literature [131, 132, 133, 78, 134, 135, 136, 137, 138,
79].
6.2.3.2 Combined heat and power node (chp)
This node represents a type of cogeneration plant where the main outputs are heat and
power. This node provides heat to the facilities in the EIN as well as potential power supply.
This node oxidizes carbon monoxide and hydrogen gases coming from the pressure swing
adsorption (psa), gasification (g) and carbon capture (cc) nodes. The overall interactions
of this node are shown in Figure 6.3.
The total amount of materials entering this node can be estimated as follows:
CIm = POm,j +GOm,j + CCOm,j +MOm,j ; j = chp (6.14)
131
Figure 6.3: Combined heat and power process
where CIm represents the total amount of material m entering the combined heat and
power unit whereas POmj, CCOm,j and MOm,j are the material outputs from pressure
swing adsorption, carbon capture and an outside source, the market, going into the chp
node, respectively (i.e., the market can provide H2 if needed). Moreover, only carbon
dioxide and water are generated in this process. The amount of carbon dioxide produced
in this node is calculated as follows:
CPM2 =∑
m∈M1,M8
CIm (6.15)
where CPm is the total amount of carbon dioxide (m = M2) produced in this facility.
The carbon monoxide and methane entering this node participate in a series of reactions
producing carbon dioxide on a 1:1 mole ratio according to standard combustion stoichiom-
etry. Additionally, water is generated in this process as follows:
132
CPM7 = CIM3 + 2CIM8 (6.16)
where CPm is the total amount of water (m = M7) produced in the combined heat and
power node, which results from reactions involving hydrogen and methane. Accordingly,
for each mole of hydrogen and methane that react in this process, one and two moles of
water are produced, respectively, according to standard combustion stoichiometry for di-
atomic hydrogen (H2) and methane (CH4).
The main material outputs associated with this process are carbon dioxide and water.
Nevertheless, nitrogen may enter and exit this node unchanged or may react to produce
NOx under some conditions. The rest of the materials that enter this node are consumed
in the internal process reactions. The materials from this node (stack gas) are sent to
carbon capture; these output materials can be estimated as follows:
COm,j = CIm + CPm ∀m ∈ M2,M4,M7 , j = cc (6.17)
where COm,j is the amount of material m sent to the carbon capture node; this product
is mainly composed of CO2. The total energy content within the gas fed to this node can
be calculated as follows:
CEC =∑
m∈M1,M2,M3,M4
CImLHVm (6.18)
where CEC is the total energy content of the gas entering the combined heat and
power unit, and LHVm is the lower heating value of the gaseous components, which are
widely published values and are included in the appendix. Furthermore, the total amount
of electricity and heat generated can be estimated as shown in Eq. 6.19 below:
133
CGu = CERuCEC CEFu ∀u ∈ U1, U3 (6.19)
where CGu is the total amount of utility u generated in the process, CERu is the
ratio of the total energy content used to produce utility type u, CEFu is a parameter that
represents the efficiency of generating utility type u. The process modeling in this section
is drawn from several major publications [131, 139, 136, 140, 141].
6.2.3.3 Carbon capture node (cc)
Captured CO2 can be purified and used for a wide variety of processes in order to avoid
emitting it to the atmosphere. Presently, monoethanolamine (MEA) is the most common
CO2 capture media used in the industrial sector. However, ammonia represents an alter-
native medium and has demonstrated various advantages in its ability to capture carbon
dioxide [81, 82]. Therefore, ammonia is included as the CO2 capture media in the proposed
network to avoid the need of importing MEA. The diagram of this process is shown here
as Figure 6.4.
The total amount of materials entering carbon capture can be calculated as follows:
CCIm = GOm,j + COm,j + POm,j + AOm,j ; j = cc (6.20)
where CCIm is the total amount of material m entering carbon capture, POm,j is the
outlet material from the pressure swing adsorption node into carbon capture, and AOm,j
is the output from the ammonia production node used as make-up material to maintain
the carbon capture process. The carbon capture process does not involve any chemical
reaction, in this process the CO2 product is purified for its use in other nodes, especially
the urea plant. Accordingly, the amount of CO2 product can be estimated as follows:
134
Figure 6.4: Carbon capture process
CCOm,j = CCSFjCCIm ;m = M2, j = up, sq (6.21)
where CCOm,j (m = M2) represents the output of CO2 from the unit, CCSFj is the
fractional splitting of the CO2 stream at the outlet of this unit to the sink nodes j. Note
that most of the purified carbon dioxide is sent to the urea production process whereas
the remaining CO2 is sent to the outlet node sq (i.e., carbon sequestration). Nevertheless,
small amounts of CO2 are also sent to the chp and psa nodes. The amount of the remaining
carbon capture outlet materials can be estimated as follows:
CCOm,j = CCSFjCCIm ∀m ∈ M1,M2,M3,M4 , j = chp, psa (6.22)
where CCSFj is the split factor of the CO2 deficient outlet gas sent to the sink nodes
j. Furthermore, the utility demands associated to the carbon capture node are electricity,
135
water and heat. The utility demands of the carbon capture facility can be estimated as
follows:
CCDu = CCRuCCOm,j ;m = M2, j = up (6.23)
where CCDu represents the demand of utility type u in the carbon capture node, and
CCRu is a parameter denoting the utility requirements per unit of CO2 product sent to
the urea plant obtained from the literature [142, 143, 144, 145, 110, 146, 147, 148, 81, 139,
149, 150, 151, 138, 152, 82, 153]. Moreover, the minimum amount of CO2 sent to the urea
plant, which depends on the urea plant reactions, is constrained as follows:
CCOM2,up ≥ 1.4 (0.5AOM5,up) (6.24)
where CCOM2,up represents the amount of CO2 sent to the urea plant and AOM5,up
is the amount of ammonia sent from the ammonia to the urea plant. This constraint
specifies the minimum amount of CO2 required for the main reaction in the urea plant.
The design flowrate of CO2 is set to be 40% greater than the stoichiometric requirement
as recommended by Riegel and Kent [117].
6.2.3.4 Pressure Swing Adsorption node (psa)
This process is typically used to separate one or more gas species from a mixture, the inlet
gas is passed over an adsorbent which attracts the desired gas or impurity in the stream,
whereas the remainder of the feed continues to the outlet for release or further processing.
This process will be used in the EIN to separate hydrogen from a gaseous mixture [86, 85].
The schematic of this process is shown in Figure 6.5.
The total amount of materials entering the pressure swing adsorption node can be
calculated as follows:
136
Figure 6.5: Pressure swing adsorption process
PIm = GOm,j + CCOm,j + AOm,j ; j = psa (6.25)
where PIm is the total amount of material m entering the pressure swing adsorption
unit, and AOm,j (j = psa) represents the outlet materials from the ammonia plant into
the psa node. The amount of hydrogen produced in this unit can be estimated as follows:
POm,j = PSFjPIm ;m = M3, j ∈ ap,mk (6.26)
where POm,j represents the amount of hydrogen product obtained in this node, PSFj
is the H2 stream splitting factor from this unit to the ammonia plant (ap) and market
(mk). However, traces of H2 products are also sent to the chp and cc nodes. Accordingly,
the remaining material outputs can be calculated as follows:
POm,j = PSjPIm ∀m ∈ M1,M2,M3,M4 , j = chp, cc (6.27)
where PSj is the split factor associated with the H2 deficient syngas sent to the pro-
duction nodes j. The PSA process is well-defined in the literature [154, 155, 138, 156] and
is used throughout industry for separating hydrogen from mixed gas streams.
137
The utility demands of the pressure swing adsorption node are electricity, water and
heat. The utility demands consider in this production node can be calculated as follows:
PDu = PRu
∑j∈ap,mk
∑m=M3
POm,j (6.28)
where PDu is the total amount of utility u required in the pressure swing adsorption
node, and PRu is a parameter that defines the energy requirements per unit of H2 product
sent to the ammonia plant and the market.
6.2.3.5 Sequestration node (sq)
Part of the objective of this model consists of minimizing waste production. This is achieved
using waste as a co-product in as many applications as possible. However, waste can only
be utilized to a certain extent, after which sequestration plays a key role in balancing the
excess of CO2 produced in the network; thus, venting gas to the atmosphere (which would
contradict part of the models objective) can be avoided.
The total amount of materials entering the sequestration (SIm) node can be estimated
as follows:
SIm = CCOm,j ;m = M2, j = sq (6.29)
where SIm (molh−1) represents the total amount of CO2 entering the sequestration
node from the carbon capture facility. The carbon dioxide sent to this node is stored in
suitable sites, e.g., deep saline aquifers. Therefore, this production node does not contain
any outlet stream.
The utility demands associated to the sequestration node can be defined as follows:
138
SDu = SRuSIm (6.30)
where SDu is the total amount of the utility type u required in the sequestration pro-
cess, and SRu is a models parameter that represents the energy requirements per unit of
CO2 entering the sequestration node as found in the literature [157, 145, 86].
The addition of carbon capture and storage represents a benefit to society in terms
of reduced emissions, yet there is an economic cost of this addition. Nevertheless, this
benefit will be more clearly quantifiable to plant operators in the conceivable circumstance
that a carbon cap and trade system or carbon taxes are implemented in order to mitigate
climate change. The literature discusses many economic and environmental aspects of
carbon sequestration in conjunction with the technical specifications of such facilities [158,
159, 150, 145, 160, 144, 157, 138].
6.2.3.6 Ammonia node (ap)
The production of ammonia in the eco-park is considered to be similar to the commonly
used Haber-Bosch process, where three molecules of hydrogen react with one molecule
of nitrogen over a catalyst to produce two molecules of anhydrous ammonia (NH3) as
described by Riegel and Kent [117]. The diagram of this process is shown as Figure 6.6.
However, the production of ammonia in this network is simpler since the feed gases are
already free of sulfur compounds and other impurities. Furthermore, this production node
may also take in unreacted ammonia from the urea plant, depending on process conditions
and geographical locations [89]. The total amount of materials entering this node can be
estimated as follows:
AIm = POm,j +MOm,j ; j = ap (6.31)
139
Figure 6.6: Ammonia process
where AIm is the total amount of material m entering the ammonia production node,
whereas POm,j, URm,j and MOm,j (j = ap) are the output materials from the pressure
swing adsorption plant, urea plant and market into the ap node, respectively. The amount
of ammonia produced in this facility can be calculated as follows:
APM5 = ACF
(2
3AIM3
)(6.32)
where APm (m = M5) is the total amount of ammonia produced in this node, ACF is
a parameter of the model that defines the total conversion of the reaction (e.g., 94%). The
amount of NH3 product out of this node can be calculated as follows:
AOm,j = ASFj (AIm + ASFmAPm) ;m = M5, j ∈ up,mk, cc (6.33)
where AOm,j represents the amount of product m in the outlet streams transported to
node j, ASFj is the ammonia plant splitting factor to nodes j, ASFm is a parameter of
140
the model that specifies the stoichiometric relationship for the production of ammonia in
this node (i.e.,ASFM5 = 1.0). The output of the reactants and remaining materials can
be estimated as follows:
AOm,j = AIm + ASFmAPM5 ∀m ∈ M3,M4 , j = psa (6.34)
where ASFm denotes the stoichiometric relationship between the amount of reactant
m consumed and the amount of ammonia produced (i.e.,ASFM5 = −1.5, ASFM4 = −0.5).
The utility demands considered in the ammonia process are as follows:
ADu = ARu
∑j∈up,mk,cc
AOm,j ;m = M5 (6.35)
where ADu is the total amount of the utility type u required in the ammonia process,
and ARu represents the parameter that denotes the utility requirements per unit of NH3
produced in this node. Ammonia manufacturing is a mature process with the current
technology and is well-documented in the literature [84, 81, 89, 117, 161, 138, 141, 152, 82,
162].
6.2.3.7 Urea node (up)
The urea production in the eco-park network is considered through the Bosch-Meiser pro-
cess where two molecules of ammonia are combined with one molecule of carbon dioxide
over a catalyst as described in the literature [84, 117]. The ammonia is readily available as
a co-product of the network and the carbon dioxide is obtained from the carbon capture
process. The process is shown pictorially in Figure 6.7.
One of the main network advantages is the energy savings due to avoiding CO2 com-
pression inefficiencies for transport since CO2 and NH3 can both be obtained from inside
141
Figure 6.7: Urea process
the EIN boundaries. The total amount of materials entering the urea production process
can be calculated as follows:
UIm = CCOm,j + AOm,j ; j = up (6.36)
where UIm is the total amount of material m entering the urea production process.
The amount of urea produced in this node can be calculated as follows:
UPM6 = UCF
(1
2UIM5
)(6.37)
where UPm (m = M6) is the total amount of urea produced in the process, UCF is a
models parameter that defines the assumed total conversion of the reaction (e.g., 85% as
described by Riegel and Kent [117]). The full process of urea manufacturing by the Bosch-
Meiser process is presented in numerous sources [84, 163, 117, 164, 138]. The amount of
urea products and any remaining co-products obtained from this node can be calculated
as follows:
UOm,j = UIm + USFm UPM6 ∀m ∈ M2,M6 , j = mk (6.38)
where UOm,j represents the amount of product m sent to the market node, USFm is a
parameter that determines the stoichiometric relationship for the production/consumption
142
of material m in this node (i.e.,USFM2 = −1.0, USFM6 = 1.0). Furthermore, the amount
of unreacted ammonia recycled within the urea node (URm,j) can be estimated as follows:
URm,j = UIm + USFm UPM6 ;m = M5, j = up (6.39)
where USFm denotes the stoichiometric relationship between the amount of reactant
m consumed and urea produced (i.e., USFM5 = −2.0). The utility demands associated
with the urea node are as follows:
UDu = URuUOm,j ;m = M6 , j = mk (6.40)
where UDu is the total amount of the utility type u required by the urea production
process, and URu is a parameter that denotes the energy requirements in the process per
unit of urea produced in this node.
6.2.3.8 Market node (mk)
The market node is not present in the network schematic presented as Figure 6.1 as it is
assumed to be connected to every node and serves a dual function in the network. Firstly,
it is the end node where the network’s products are sold to generate profits; however, it
can also act as an external supplier to help meet the material and energy needs associated
to the network when required. Accordingly, the total amount of materials entering the
market node can be calculated as follows:
MIm = UOm,j + AOm,j + POm,j ; j = mk (6.41)
where MIm is the total amount of material m entering the market node. The type and
amount of utility sold to the market can be estimated as:
143
MEIu = MERuCGu ∀u ∈ U1, U3 (6.42)
where MEIu is the total amount of utility type u sold to the market from the chp node
and MERu represents the ratio of utility u that can be sold to the market. The amount
and type of materials sent from the market to the EIN nodes are as follows:
MOm,j = MAMm,j (6.43)
where MOm,j is the total amount of material m out of the market to node j, and
MAMm,j represents the type and amount of material m available in the market to be sold
to the networks node j. The type and amount of utility sold to the networks nodes can be
estimated as:
MEOu,j = MEAu,j (6.44)
where MEOu,j is the total amount of utility type u sold to the networks node and
MEAu,j represents the amount of utility u that can be exported from the market to
supply the eco-park nodes.
6.2.3.9 Waste and Water Treatment node (wt)
The water treatment plant is used to process the water consumed in the EIN. The total
amount of water demanded by the network can be calculated as follows:
WDu = GDu + CCDu + PDu + SDu + ADu + UDu − CPM7 ;u = U2 (6.45)
144
where WDu (u = U2) is the total amount of process water demanded by the eco-
industrial network and treated in the plant. Moreover, heat and electricity are also de-
manded for the operation of the water treatment plant. Accordingly, the amount of utility
u required to treat the process water consumed in the EIN can be calculated as follows:
WDu = WRuWDu2 ∀u ∈ U1, U3 (6.46)
where WDu represents the amount of utility type u consumed in the water treatment
plant and WRu denotes the energy requirements per unit of process water produced.
On the other hand, the amount of materials treated in the waste treatment facility of this
node can be calculated as follows:
WIm = GWm (6.47)
where WIm represents the amount of materials m entering the waste treatment unit,
whereas, GWm is the amount of waste materials (i.e., ash, sulfur and other contaminants)
sent from the gasification node to waste treatment. Industrial water and water treatment
is discussed in detail within industrial literature and is implemented here based on several
sources [165, 166, 167, 168, 169, 170, 138, 171, 172, 173].
6.2.3.10 Transportation system
The transportation system enables the carriage of materials from node i to j; thus, linking
the facilities located inside the EIN. These materials can be solid, liquid or gaseous depend-
ing on the operating or storage conditions. As a result, different transportation options,
k, are included in the model to meet the specifications of the transported materials inside
the network. The transportation system is constrained to select only one option between
nodes; this can be formulated as follows:
145
yi,j,k =
1 if transportation method k is used between node i and node j
0 Otherwise
(6.48)
∑k
yi,j,k ≤ 1 (6.49)
Eq. 6.49 determines the transportation option selected to transfer materials between
nodes i and j and stipulates that only one of these methods can be used. The material
supply constraint from node i to j can be expressed as follows:
∑k
TUNi,j,kTCk ≥∑m
MATERIALm,i,j (6.50)
where TUNi,j,k is an integer variable that represents the number of transport units of
type k are required to transfer materials from i to j, TCk is the capacity of the trans-
portation method k, and MATERIALm,i,j is a variable that represents the network nodes
sending material to node j.
6.2.3.11 Network Costs
This section describes the different costs involved in the operation of the EIN such as the
transportation system cost, capital cost of the plants (nodes), the plant operating costs
and life-cycle emissions costs. Accordingly, the costs of the transportation systems are
presented forthwith. The cost associated to the pipelines transportation system can be
calculated as follows:
NTCk =∑i
∑j
(TUNi,j,kFCk + yi,j,kdi,jV Ck
∑m
MATERIALm,i,j
)∀k ∈ P (6.51)
146
where NTCk represents the networks transport cost, FCk is the fixed cost associated
to transportation method k, di,j is the distance between nodes i and j, and V Ck is the
variable cost of the transportation method k. Similarly, the transportation costs associated
to the rail and road infrastructures can be estimated as follows:
NTCk =∑i
∑j
(yi,j,kdi,jFCk + TUNi,j,k (FCUk + TCkV Ck))∀k ∈ Rd ∪Rl (6.52)
where FCUk represents the systems fixed cost per transportation unit type k. Costing
for transportation options can be found in the literature [29, 174, 175, 176].
The present model compares the economics of using integrated production plants in-
stead of stand-alone plants. Consequently, the capital costs of the plants are calculated as
follows:
xi,l =
1 if plant i of size l is selected
0 Otherwise(6.53)
ACCi,s = ACFi
(∑l
xi,lPCi,l,s +∑k
NTCk
)(6.54)
where the index s, as in s ∈ S = a, b represents the production scheme of the plant,
which can be either stand-alone (a) or integrated (b), ACCi,s is the annual amortized
capital cost of the plant i with scheme s, ACFi is the amortized capital factor of plant i,
and PCl,i,s is the capital cost of the plant. Also, the operating costs of the plants can be
calculated as follows:
OCi,s =OCFi,sACCi,s
ACFi(6.55)
147
where OCi,s represents the total operating cost of plant i following scheme s, and OCFi,s
is the operating cost factor associated to the plant as calculated from the literature.
The supply constraints for the production plants can be expressed as follows:
∑l
xi,lPICi,l ≥∑m
∑j
MATERIALm,i,j (6.56)
∑l
xi,l ≤ 1 (6.57)
where PPCl,i represents the plants production capacity.
6.2.3.12 Network Emissions
An analysis regarding the lifecycle emissions of the production nodes considered in the EIN
is included in the present model. Accordingly, the emissions associated to the construction
of the production plants included in the network can be estimated as follows:
EPCi,s,e =∑l
xi,lEFPCi,l,s,e (6.58)
where EPCi,s,e represents the emission e related to the construction of plant i of scheme
s, and EFPCl,i,s,e is a factor denoting the emission e from the construction of a plant of
size l as found in the literature for each node. Similarly, the emissions associated to the
operation of the production nodes can be calculated as follows:
EPOi,s,e = (EFPOi,s,ePOCi,s)POLi (6.59)
where EPOi,s,e represents the emission e from the operation of plant i of scheme s,
EFPOi,s,e denotes the emission factor related to the operation of the plants which are
148
found in the literature for each node, POCi,s is the plants operating capacity, and POLi
is the plants operating life (e.g., 30 years).
6.2.3.13 Objective Function
The model’s objective function is formulated in terms of two ratios: The first ratio denotes
the relation between construction and operating costs (i.e., production costs) of the inte-
grated scheme and stand-alone plants; whereas, the second ratio indicates their relative
lifecycle emissions. Accordingly, the total production cost of the network can be calculated
as follows:
NPC =
∑s=b
∑i
ACCi,s +OCi,s∑s=a
∑i
ACCi,s +OCi,s(6.60)
where NPC represents the EIN annual production cost, while the numerator and de-
nominator terms of the equation denote the production costs related to the integrated and
stand-alone schemes, respectively. Similarly, the life-cycle emissions can be calculated as
follows:
NEC =1
Ne
∑i
∑s=b
∑e
EPCi,s,e + EPOi,s,e∑i
∑s=a
∑e
EPCi,s,e + EPOi,s,e
(6.61)
where NEC represents the annual lifecycle emission associated to the network. The
reciprocal Ne factor in this equation is representative of the number of emissions considered
in the analysis so as to reduce the weight of the environmental objective to align with the
economic portion. Consequently, the objective function can be formulated as follows:
CF = WP (NPC) +WE (NEC) (6.62)
149
where CF represents the objective cost function of the problem, WP is the weight as-
signed to the profitability of the network, whereas WE is the weight considered for the
lifecycle emissions of the eco-park. The objective function takes into account the costs
related to the operation of the network as well as associated life-cycle emissions (i.e., en-
vironmental and social impacts). This work clearly shows that emission reductions for
society and improved profits for operators can be achieved through the implementation of
eco-industrial networks.
The fractional output shows a relativistic result with reference to the base case. Ac-
cordingly, if the ratio is greater than unity, the integrated scenario costs more than the
stand-alone, whereas for a ratio less than unity, it is less costly. Likewise, if the emissions
ratio is greater than unity, the integrated scenario emissions are higher than the stand-alone
scenario. On the other hand, for emissions ratio less than unity, the integrated scenario
emissions are lower than the stand-alone. The model’s objective function has been formu-
lated in terms of these ratios to assign relevance to the environmental impacts caused by
the network operations. This prevents the economic term of the objective function from
becoming very large compared to the life-cycle emission term, which could negate the life-
cycle emissions effect on the design of the network (as occurs in many cases). As a result,
both terms of the objective function play a key role in model formulation. Accordingly,
when one separates and normalizes the life-cycle emissions from the economics, one of the
terms would have to be significantly larger in order to overshadow the effect of the other
term, since they are normalized to the stand-alone emissions/economics.
6.3 Results and Discussion
The model was constructed with the basis of a minimum production of hydrogen to meet
demand for 1000 fuel cell vehicles in the province of Ontario, Canada as was the require-
150
ment the previous LP and MILP formulations of this case study presented in previous
chapters.
Previous work using LP and MILP models calculated a reduction in emissions for the
base scenario of 50% weighting on each portion of the objective to be approximately 40%
for the LP model and 35% for the MILP model [69, 108]. The work herein is updated to
reflect current pricing, introduces additional accuracy of the model and a more consistent
data set for calculating the life-cycle emissions. In addition, the model is expanded into
the MINLP domain to allow for additional constraints to be placed on the model which
were not possible using linear programming methods.
For this work, an upper limit on the input biomass feedstock was set to 2 million
tonnes per year as a sustainable rate for harvesting from forest residues and other undesir-
able biomass co-products [177]. No limits were placed on coal or natural gas availability.
The plant lifetimes were assumed to be 30 years as is a standard assumption in chemical
engineering and the capital charge rate for amortization was set to be 15%. The following
scenarios discussed are based on varying the weighting factors for the environmental and
economic portions of the objective function from 0 to 1 by increments of 0.1. The purpose
of this increment is to explore the different network configurations and outputs with varying
priorities of the network management. The complexity of the network demands significant
computing resources and it was determined that a smaller increment would require exces-
sive computing resources and that scoping on the coarser level would exhibit the variation
of the network outputs with respect to the economic and environmental weighting factors.
The reference sizing of each facility is shown in Table 6.1. The scaling for each facility
was completed using the capacity scaling equation used by many chemical plant design
151
texts and is shown as Eq. 6.63.
PlantCost2 = PlantCost1
(PlantCapacity2
PlantCapacity1
)q(6.63)
The scaling factor, q, was assumed to be 0.6 for these facilities as is a generally accepted
practice in the industry. Chemical process equipment is typically subject to sizing options
given by the manufacturer and thus the plant sizes available were chosen as even multiples
of the reference plant size (up to a multiple of five) or even fractions such as 14
, 12
or 34
of
the reference condition.
The values of the weighting factors in the bi-objective function are varied from 0 to 1
in an attempt to analyze the difference in facility construction with emphasis placed on
economics, reduction in emissions or a combination of the two. With a purely economic ob-
jective function, the resulting network is shown in Figure 6.8 with the major mass flowrates
shown in Table 6.2 which is designated by the number above each flow stream. It can be
seen from this that ammonia production in the network is foregone completely in favour
of purchasing ammonia from outside of the eco-park. Since a price premium is not applied
to importing ammonia, it is reasonable that ammonia would be imported to the network
under a purely economic scenario. In addition to this, carbon sequestration is also not
included in the network. Since the objective in this case is completely economic, and
viable products are not produced from carbon sequestration, it is again reasonable that
this facility would not be constructed. The material exports from the network under the
purely economic case are then limited to hydrogen and urea. Urea is a value-added product
compared to the inlet ammonia and CO2 streams; thus, it is maintained in the network
where ammonia production is excluded. The air separation for gasification would also
lead to a saleable nitrogen product in this case, although this was not considered as part
of the economic analysis as the nitrogen is simply used as a reagent in ammonia production.
152
Table 6.1: Capacities and reference costs of eco-park facilities
Unit Reference Capacity Reference cost
($millions)
Literature Reference
Gasifier - Coal (TPD coal
feed)
3 390 269 [134]
Gasifier - Biomass TPD
biomass feed
1 130 134 [136]
Gasification (Steam
Methane Reforming)
(TPD natural gas)
3 120 255 [178]
Carbon Capture (TPD cap-
tured CO2)
11 000 547 [126]
Combined Heat and Power
(MWe)
335 208 [179, 134]
Pressure Swing Adsorption
(TPD H2)
229 10 [180]
Sequestration (TPD CO2
sequestered)
2 690 114 [158]
Ammonia Production
(TPD ammonia produced)
1 800 678 [181]
Urea Production (TPD urea
produced)
2 350 189 [182]
Water Treatment (TPD
treated water)
9 450 3.16 [183]
153
Figure 6.8: Optimal network schematic under purely economic consideration
The same analysis has been conducted over a range of values for the objective func-
tion weights, the case in which the environmental objective term dominates the objective
function, WE = 1, is shown in Figure 6.9 with the accompanying flowrates shown in Table
6.3. The purely environmental objective shown here includes smaller facilities, yet indi-
cates that the facilities are all included in the network, despite the environmental burden
associated with their construction. The level of production at each facility is less than
shown in alternative scenarios and it was necessary to include a positive production con-
straint from the gasification unit in order for the optimization to solve without a trivial
solution for this particular scenario. CO2 is sequestered in this scenario, though at a lower
rate than observed in other scenarios during this ten-scenario scoping analysis. This is
due to the large energy burden required for CO2 sequestration and the already-reduced
plant sizing. Consequently, the additional energy required to sequester additional CO2
would be provided by the external electricity grid (i.e., adding energy costs). Hydrogen in
154
Table 6.2: Summary of flowrates under pure economic consideration
Line number Description Million Tonnes per annum
0 Gasifier Outlet 2.77
1 CHP Outlet 2.61
2 CO2 Sequestered N/A
3 CO2 to Urea production 0.429
4 Hydrogen to ammonia manufacturing N/A
5 Hydrogen to combustion 0.0547
6 Ammonia to Urea production N/A
7 Ammonia exported N/A
8 Urea exported 0.363
this scenario’s network is not used for combustion as was observed in some other scenarios;
however, the hydrogen is utilized for producing ammonia which in turn is used as a reagent
for producing the urea, sequestering CO2 by proxy.
Figure 6.10 shows the variation in the objective value varying the weights of the two
parts of the bi-objective function described by Eq. 6.62. This figure shows the balance
between the objectives for a range of solutions to the problem at set parameters. In the
case of a bi-objective function in which the sum of the weighing factors is equal to unity,
the objective value can be shown as a correlation with only one of the weighting factors;
this two-dimensional view is shown as Figure 6.11.
Figure 6.10 and Figure 6.11 show the variation in the objective value based on a vari-
ation of the two terms of the objective function. The objective function is based on
environmental and economic benefits obtained from operating in an integrated EIN com-
155
Figure 6.9: Optimal network schematic under purely environmental consideration
Table 6.3: Summary of flowrates under purely environmental consideration
Line Description Million Tonnes per annum
0 Gasifier Outlet 0.264
1 CHP Outlet 3.39
2 CO2 Sequestered 0.764
3 CO2 to Urea production 0.109
4 Hydrogen to ammonia manufacturing 0.030
5 Hydrogen to combustion N/A
6 Ammonia to Urea production 0.132
7 Ammonia exported 0.129
8 Urea exported 0.282
156
Figure 6.10: Results of the objective value for eco-park optimization varying objective
function weights
pared to facilities operating independently. These results show that the best objective
value is reached when the life-cycle emissions weighting is set to zero. The optimization
at this point is essentially an economic optimization and shows that there are significant
economic incentives to be considered for operating as part of an integrated network of
facilities. Each of the scenarios presented in the figures are independent and should not
be compared in order to determine which scenario is ultimately the best. In this case,
however, it is notable that the weighting of the two objectives is highly influential upon
the objective value found by the optimization. The optimization results show that the
economic term is more favourable to a minimum objective than is an increased weighting
upon the life-cycle emissions.
157
Figure 6.11: Results of variation in the objective value weights in two dimensions related
to weighting for life-cycle emissions
The size of each facility in the network is allowed to vary in order to obtain the optimal
objective value within the simulation and thus it is interesting to include the normalized
capacity of the facilities for each of the scenarios for which the objective function weights
were varied. The result is essentially a sensitivity analysis on the weighting parameters
in the bi-objective function. The solution time for the MINLP is typically 7 - 10 days and
thus additional sensitivity analysis on other parameters was not completed. The normal-
ized plant capacities from varying the weighting factors in the objective function are shown
in Figure 6.12.
The gasification node shows a maximum production for the economic mono-objective.
A sharp decline in gasifier usage is shown with the introduction of the emissions portion of
the objective function. Since the gasifier is a relatively large contributor to the emissions of
all contaminants considered in this work, it becomes obvious that its size would be reduced
158
Figure 6.12: Normalized plant capacities under varying weighting of life-cycle emissions.
G is descriptive of gasification, CC of carbon capture, CHP for combined heat and power,
PSA for pressure-swing adsorption, SQ for sequestration, AM for ammonia manufacture
and U for urea production
to a much lower level once the network emissions become incorporated into the objective
function via the life-cycle emissions weighting factor.
Carbon capture processing within the network is intended to extract CO2 from exhaust
gases for purposes of sequestration or utilization in further reactions. For most of the sce-
narios, carbon capture processes remain at a relatively low level relative to the maximum
available plant capacity. In many cases, it is favourable to purchase CO2 from the mar-
ket to avoid the contaminants associated with producing the CO2 onsite. The exception
to this is during the instances in which the combined heat and power unit operates at a
high level, requiring the carbon capture capacity to increase to balance the loading from
this unit. Indeed, the maximum carbon capture capacity occurs in conjunction with the
159
maximum output from the CHP unit and also in the case of a maximum CO2 sequestration.
The combined heat and power unit supplies heat and/or electricity to the processes
within the network and excess of either product is exported through the market node. The
variability in CHP output is directly related to the increased ammonia production and is
favourable under only two of the simulations considered. Accordingly, a higher production
of ammonia demands more heat and electricity, which causes an increase in the CHP unit
size to accommodate the production of ammonia.
Hydrogen is separated in the PSA unit and utilized as a feed to the other network
nodes but can also be exported as a market product. The level of production for the PSA
has a lower limit set to produce hydrogen for a minimum of 1000 fuel cell vehicles in the
province of Ontario, as in the previous analysis, but there is additional incentive under the
objective function herein to produce additional hydrogen. This level of hydrogen produc-
tion was selected as a likely entry point when dedicated facilities will be required to enable
the hydrogen economy. The benefits of producing hydrogen using the proposed network of
facilities are both economic and environmental; therefore, hydrogen production is a neces-
sity to reduce the emissions burden from further processes but can also be produced as a
reduced-emissions export.
Carbon sequestration is applied when the economic and environmental objective would
be better served by its inclusion than its exclusion. The economics of sequestration are
undesirable as it lacks a market value under the current pricing regime; therefore, the
inclusion of carbon sequestration processing in the network is related solely to the environ-
mental benefits. In the majority of simulations, sequestration is excluded or constructed
on a very small scale. The anomalous occurrences are found when the life-cycle emissions
160
weighting factor is at 0.4 or 0.7 and complement a larger burden from the gasification unit.
Ammonia production, contrary to the results of previous work, is maintained at a rel-
atively low level in most simulations presented here. The juxtaposition of ammonia and
urea production is a direct correlate to the price of these two products which was not the
case in the previous published work [69, 108]. This exhibits the leverage of product cost
on the simulation as urea is a more profitable product in this analysis and as the urea
node approaches the maximum processing capacity, ammonia is sold to the market as an
additional method for increasing the economics of the EIN.
Urea production remains relatively constant throughout the simulations, having only
a slight decline to 75% of its maximum value for the environmental mono-objective. The
difference between the production of urea under these simulations compared to the results
presented in previous chapters is the increment in the urea price. Additionally, the overall
objective function differs in the environmental objective term, which compares release of
GHG emissions in an integrated and stand-alone facility concurrently. In previous publi-
cations using LP and MILP modeling, urea production was generally low which could be
attributed to its relatively low reduction in absolute GHG emissions. With the objective
function as implemented in this work, the relative reduction of GHG emissions relative to
a stand-alone urea facility are much higher, corresponding to relatively stable levels of urea
production.
Another consideration of the model solution is which part of the bi-objective dominates
the result. In Figure 6.13 and Figure 6.14, the results of this question are addressed. The
scalar weighting factor of each portion of the objective function is applied to the objective
value resulting from the optimization to evaluate the impact of each portion of the objective
161
Figure 6.13: Relative contributions of emission reduction objective and economic objective
to the overall objective value considering weighting factors
function on its final value. The contribution from each portion of the objective function is
comparable to the weighting applied, though the emissions term in the objective function
contributes a marginal surplus above the specified weighting factor for all cases except the
mono-objective considerations of WE = 0, 1. Figure 6.13 includes the weighting factors for
each part of the bi-objective function while Figure 6.14 shows the relative weight of each
part of the objective function irrespective of the weighting applied during the optimization.
This result is also shown by the contribution of each portion of the objective function
excluding the weighting factors as shown in Figure 6.14. This figure exhibits the relative
strength of the two components of the objective function in each of the scenarios consid-
ered. This reinforces the finding that the objective value for the life-cycle emissions is
slightly more influential on the final objective value than is the impact of the economic
portion of the objective function.
162
Figure 6.14: Relative strength of emission reduction and cost reduction factors on the
objective function
This work and the previous chapters differ in terms of the reduction in cost and emis-
sions. For the same base case as in previous models, the emissions are reduced by 12.7%
and the cost savings amount to 24.6% which are more reasonable and conservative esti-
mates compared with the previous model results. The LP and MILP formulations of this
case study presented in earlier publications should be considered as an upper boundary for
the reduction in cost or emissions from operating in an integrated network of facilities as
the full complexity of the system cannot be captured comprehensively by these methods.
The MINLP model is a more realistic assessment of the actual cost and emission reductions
from such operating an EIN.
163
6.4 Summary
This chapter presents an MINLP modeling framework for assessing new construction of an
EIN with many integrated nodes. One major difference between the model in this chap-
ter and those in the previous two chapters is the relativistic bi-objective function which
optmizes the configuration of the network based on percentage reduction in each emission
and percentage of improvement in overall economics. This relationship is non-linear and
thus cannot be applied in either the LP or MILP formulations in Chapters 4 and 5. This
novel approach to EIN design is a powerful tool for policy-makers to understand the impli-
cations of regulation and pollution abatement relative to the status quo. This framework
allows for quantitative discussions between policy-makers and industry with the intention
of improving the environmental performance of chemical processes and introducing cost
savings potential for improved profitability.
The solution to the MINLP shows high variability in the capacity of various facilities
in the EIN relative to the weighting of the two parts of the bi-objective function. Con-
trasting the previous two chapters, each combination of weighting factors yields different
plant capacities and configurations of the network. The model is highly sensitive to these
weighting factors and it is likely that parameters such as product pricing would also have
an impact on the optimal solution /network configuration.
The opportunity for collaboration between chemical processors and with regulators
negates the thinking that reducing emissions and improving profitability are mutually
exclusive. The framework developed here with the bi-objective function comprised of an
economic term and environmental term relative to existing operations shows that both
can be improved without compromising the other. While some scenarios show greater
profitability or a greater reduction in emissions, policy-makers can participate in a realistic
164
Chapter 7
Summary and Conclusions
7.1 Summary of Work
This work developed several models of an eco-park in order to demonstrate the environ-
mental and economic benefits of industrial facilities working cooperatively to share energy
and materials, including the production of hydrogen to support a future hydrogen economy.
The EIN models constructed in this work focus on cradle-to-gate emissions from energy
conversion and the production of industrial chemicals. This work utilizes three distinct
modeling techniques in order to demonstrate their respective benefits, specifically: linear
programming, mixed-integer linear programming and mixed-integer non-linear program-
ming. Chapter 3 in this thesis addresses life-cycle emissions of criteria air contaminants
from vehicle operations and potential reductions that could be experienced by shifting to
alternative vehicle fuels. The work presented in Chapter 3 is focused on the province-wide
emissions in Ontario, Canada and urban air pollution in the city of Toronto.
Chapter 4 expands on the utilization of life-cycle concepts but is focused in the ap-
plication of the methodology to an EIN. An optimization method is developed in order
166
to balance the frequently-conflicting goals of reduced cost and reduced emissions of con-
taminants to the air, water and land. This chapter explored eco-industrial integration
using an LP model and revealed some preliminary conclusions. From this analysis, it was
found that hydrogen can feasibly be produced to fuel 1000 hydrogen vehicles in Ontario
within the proposed EIN. Policy-makers are the intended audience for this approach in
addition to forward-thinking facility designers and environmental organizations. Utilizing
this modeling framework to negotiate an acceptable plan for facility construction based on
non-zero weighting factors for economic and environmental decisions allows for a quantita-
tive discussion of benefits for policy-makers and industry. The focus on production in an
EIN complements the vehicle-centric work from Chapter 3 which exhibits benefits beyond
the facility gate for utilization of a potential chemical fuel produced in an EIN.
Chapter 5 attempts to address these recommendations by expanding the model to an
MILP formulation and also exploring three feedstock fuels for the analysis. Chapter 5 con-
tinues the work from Chapter 4 with added complexity and realism which required adding
binary/integer decisions. This work builds on the previous work to represent a quantitative
assessment of the eco-park theory and its application to a case involving the production
and export of several chemical products in addition to heat and electricity from an EIN
comprised of a number of facilities. The network is assessed in terms of environmental
impact and profitability relative to existing facilities that do not interact directly with the
exchange of material and energy streams amongst the nodes. GAMS software was again
employed to create the optimization program with decision variables describing which of
the network nodes should be constructed and what their associated production capacities
should be in order to optimize a similar dual-objective function including profit and en-
vironmental impact. Again, the focus is to provide a balance between cost-savings and
reduced environmental impact to influence policy and the design of facilities.
167
Chapter 6 builds further on the previous modeling attempts and migrates the model
to a mixed-integer non-linear program (MINLP). This is the most detailed model of the
system constructed to date and implements a relativistic objective function to normalize
the outputs in relation to independent facilities. This objective function is a valuable tool
for policy-makers as it exhibits the increase or decrease of process profitability and emis-
sions in an integrated scenario relative to the base case of no integration. This is capable
of influencing policy decisions on facility construction in order to achieve net benefits for
producers as well as society.
7.2 Conclusions
The life-cycle impacts of utilizing alternative fuels for transportation purposes is considered
in terms of six major stressors for climate change, acidification and urban air quality. The
vehicles considered are plug-in hybrid electric vehicles (PHEVs), fuel cell vehicles (FCVs)
and fuel cell plug-in hybrid electric vehicles (FCPHEVs). Modeling of the penetration rates
for these types of vehicles has been completed based on the maximum base-load capacity
of Ontario’s electricity grid to accommodate the generation of hydrogen and charging of
vehicles using grid electricity. Results show that the reduction in greenhouse gas emissions
from adoption of PHEVs or FCVs will exceed 3% of the current emissions from the trans-
portation sector in Ontario while FCPHEVs may achieve almost twice this reduction. All
vehicles exhibit similar impacts on the precursors for photochemical smog although the
province-wide effects differ significantly.
Also from chapter 3, it is observed that the location of new generation in Ontario greatly
affects the supportable penetration of AFVs as both vehicle types depend on electricity
to generate their fuel. This study focused on near term evaluation of the potential pene-
168
tration of likely PHEV and FCV technologies, and as such considered only the currently
planned upgrades to the electrical generation system. The only scenario considered for this
study was that new generation will be located in the Toronto region. Locating generation
in the Bruce zone will decrease the supportable penetration of PHEVs mainly due to the
proximity to the urban population. Clearly if greater emission reductions in GHGs and
urban air pollutants from vehicles are desired, further large scale expansion of the CO2-free
generation sources such as nuclear will be required and this will dramatically change the
available base-load power profile.
Comparing FCVs and PHEVs in an urban-emissions scenario shows that FCVs tend
to present an advantage over or near-equality with PHEVs in almost every aspect of their
emissions. The range of FCPHEV emissions for the pertinent urban air pollutants suggest
that a minimum number of FCPHEVs would exhibit reductions close to those attained
by PHEVs or FCVs and the potential for the reduction of these emissions could greatly
exceed that achievable by either vehicle.
One of the major emissions impacts from adopting either PHEVs or FCVs is a 3-3.5%
decrease in greenhouse gas emissions within the transportation sector. Since transportation
accounts for approximately 37% of all emissions of these gases in Canada, it is a signif-
icant impact to reduce transportation GHG emissions in Ontario by 3-3.5%. Adoption
of FCPHEVs could attain a reduction as high as 5.8% of the transportation emissions in
Ontario.
The effects on emissions of particulate matter are drastically different for the adoption
of the different vehicle types. The overall emissions of particulate matter are of compar-
atively less concern because this matter is easily dispersed in the atmosphere instead of
169
being heavily concentrated in a densely-populated area where it can affect the health of
the population. Adoption of PHEVs shows an insignificant increase in overall particulate
matter whereas adoption of FCVs and PHEVs will achieve approximately the same reduc-
tion in urban particulate matter emissions. FCPHEVs perform similarly to PHEVs for
increased emissions of overall particulate matter but show a potential for larger decreases
in the urban environment.
Overall emissions of SOx, a contributor to acid rain, are expected to increase from the
adoption of PHEVs and FCPHEVs which corresponds to an increase in the burning of coal
in order to produce electricity for powering the vehicles.
When observing the emissions and technological readiness of each vehicle type, it should
be noted that PHEVs are nearer to mass production and distribution but should be cou-
pled with fuel cells as soon as production is technologically and economically feasible in
order to achieve the maximum reduction in emissions.
The work from Chapter 4 focuses on analyzing possiblities for reduced life-cycle emis-
sions from industrial operations producing industrial chemicals including hydrogen for fuel
cell vehicles. The model developed is specifically for assessing the economics and emission
reductions from EINs but the environmental benefits are limited to the boundaries of the
EIN. Reduced emissions from the production of industrial chemicals and fuels can also have
impacts throughout the remainder of the life cycle as shown by the emission reductions
from AFVs.
Applying this model for a scenario of hydrogen production for 1000 consumer vehicles
shows two independent stable solutions. Both of these solutions show profitability while
170
reducing life-cycle emissions and maintaining the appropriate production of hydrogen for
these vehicles. The LP model developed in this chapter is powerful but lacking in the com-
plexity required to fully capture the interactions between networked production facilities.
This assessment emphasizes the need for further efforts in network modeling, leading to an
MILP or MINLP model which will also allow for consideration of alternative gasification
feedstocks.
Chapter 5 shows that life cycle analysis and eco-park network optimization provide
an excellent pairing for assessing EINs. The emissions from producing typical chemical
exports are reduced when operating within the construct of an EIN and the profitability of
the network is also improved. The optimization of this network with three different fuels
shows that biomass usage has slightly less effect on the environment but with reduced profit
and a net import of electricity from an outside source. A mixed integer modeling technique
has been applied to improve potential application of the simulation results. Comparisons
of three fuel types showed that biomass yielded the least profitable scenario, required a net
import of electricity and produced significantly less heat for use in surrounding applications
as would be suggested to improve the environmental performance of the region.
Additionally, utilization of three different fuels as the primary source of chemical
reagents and energy are considered in various scenarios. These fuels are coal and biomass
for gasification or a steam-methane reforming process using natural gas as the feedstock.
The eco-park with interacting production nodes is shown to be more profitable than the
comparable non-integrated set of facilities and the outputs are produced with lower envi-
ronmental impact in terms of criteria air contaminants.
The model in Chapter 6 is a reconstruction and expansion from the MILP model in
171
Chapter 5 to reduce approximations and expand the capabilities of the model to handle
a broader range of considerations. The MINLP includes a more detailed assessment of
transportation considerations in addition to a more rigorous and complex set of contraints
governing the interactions between plants.
This model builds on previous iterations and lends itself to future development as a
tool to assess environmental and economic feasibility of applying eco-industrial integration
concepts to new construction projects. In addition, the model is formulated to assess fea-
sible transportation distances, transportation technologies and may be modified to include
stochastic import/export pricing. The results of this analysis show that the estimated cost
and emission reductions are less significant than with an LP or MILP model but represent
a more realistic case. Cost for the integrated set of facilities is shown to be reduced by
24% while the emission reduction is observed at 12.7% for the base scenario represented
by the earlier LP and MILP models.
The work from this chapter further expresses the usefulness of MINLP modeling to
assess the potential benefits of eco-industrial integration. The MINLP model is the most
comprehensive of the three considered in this work for assessing eco-park scenarios though
it requires significantly more computational power.
7.3 Summary Statement of Contributions
This work has focused on providing a model for the analysis of the improvement of the
environmental performance of transportation technologies and industrial manufacture of
base chemicals in the context of emissions to the aquatic, terrestrial and atmospheric do-
mains.
172
This work has developed a novel method for analyzing the potential and impacts of
alternative-fuel vehicles and a generalized modeling framework for eco-industrial network
integration using linear programming, mixed-integer linear programming and mixed-integer
non-linear programming. These modeling frameworks have been applied to a potential
eco-park integration scenario in order to assess the realistic potential for decreased envi-
ronmental emissions and improved process profitability. This analysis has been completed
in the broader context of the hydrogen economy and a social desire to improve the quality
of the environment. As such, the work has also demonstrated the benefits and differences
of the modeling techniques explored herein.
Life-cycle assessment has been applied to construct a dual-objective function capable
of balancing environmental concerns with industrial profitability. The objective functions
for these analyses relate the emissions and economics of independent facilities with those
operating within an eco-industrial network arrangment. The objective function in each
framework is of particular interest to policy-makers who can utilize such methods to find
a compromise with industrial entities to reduce emissions while implementing cost-saving
measures.
The bi-objective optimization framework in this thesis is a novel approach to facility
design and includes environmental and economic considerations. Emissions are considered
independently of economics in each model to avoid large economic incentives from over-
shadowing environmental objectives.
This work also serves to re-invigorate the notion of eco-industrial integration in light
of increasing corporate social responsibility and the societal shift to provide increasing
173
environmental awareness in industry. The economic needs of industry are balanced with
the societal desire for reducing emissions to achieve mutual benefits.
7.4 Recommendations
7.4.1 Recommendations for Future Work
Recommendations for further work on this topic include reformulation strategies of the
MINLP to reduce the solution time which remains one of the limitations for this type of
model. MINLP modeling is well-suited to optimization of industrial eco-parks, yet the
solution time is such that large numbers of simulations cannot be completed in a rapid
fashion.
Stochastic simulation of networks which rely heavily on static costs of base chemicals
should be conducted. Prices of both the feedstock and products of these networks may
be highly sensitive to these costs and thus it is important to explore predicted values for
commodities and chemicals.
In order to expand the list of contaminants considered for an optimization of this struc-
ture, a more comprehensive database of contaminants must be available for use. Research
and data for less widely-studied contaminants are difficult to locate and contain many
uncertainties as many aspects of production are not considered. Further work into devel-
oping a comprehensive database of life-cycle emissions must be available and be backed by
high-quality research and methods.
174
7.4.2 Recommendations for Implementation
This work exhibits the impact of adoption of alternative fuel vehicles on the air quality in
the urban centre of Toronto, Canada as well as the impact on air pollution in the province
of Ontario. It is recommended that policy-makers create incentives for transitioning from
convential fossil-fuel powered vehicles to those powered by electricity or hydrogen. The
short-term benefits can be realized immediately by transitioning the existing vehicle fleet
to operate on electricity with greater improvements seen in the long-term adaptation to
hydrogen-powered vehicles. These recommendations are within the broader context of the
hydrogen economy in an attempt to curb urban air pollution and smog in addition to global
concerns of climate change. The transitions mentioned in this work are within the scope
of electricity production in Ontario, Canada as forecast to 2025.
Based on the EIN case studies, it is recommended that industrial entities strongly
consider the use of eco-industrial design optimization in order to realize the benefits of
eco-industrial integration. Such tools would improve the economics and environmental
performance of operations, proactively making improvements to processing in the eventual
scenario of greater legislation on emissions of contaminants to air, water and land. In
addition, this work serves to instruct policy-makers to consider stricter environmental
legislation with the goal of encouraging corporations to apply eco-industrial integration
possibilities into new facility constuction.
175
References
[1] A.J.D Lambert and F.A Boons. Eco-industrial parks: stimulating sustainable devel-
opment in mixed industrial parks. Technovation, 22(8):471–484, August 2002.
[2] A.G. Matani. Strategies for better waste management in industrial estates. Journal
of Industrial Pollution Control, 22(1):67–72, 2006.
[3] Ernest A. Lowe. Creating by-product resource exchanges: Strategies for eco-
industrial parks. Journal of Cleaner Production, 5(1-2):57–65, 1997.
[4] Marian R. Chertow. Industrial symbiosis. In Encyclopedia of Energy, pages 407–415.
Elsevier, New York, 2004.
[5] N.B. Jacobsen. Industrial symbiosis in kalundborg, denmark: A quantitative as-
sessment of economic and environmental aspects. Journal of Industrial Ecology,
10(1-2):239–255, 2006.
[6] M.P. Olesen. Industrial symbiosis in kalundborg. VGB PowerTech, 79(10):52–54,
1999.
[7] Brian H Roberts. The application of industrial ecology principles and planning guide-
lines for the development of eco-industrial parks: an australian case study. Journal
of Cleaner Production, 12(8-10):997–1010, October 2004.
176
[8] Deog-Seong Oh, Kyung-Bae Kim, and Sook-Young Jeong. Eco-industrial park design:
a daedeok technovalley case study. Habitat International, 29(2):269–284, June 2005.
[9] Magnus Karlsson and Anna Wolf. Using an optimization model to evaluate the
economic benefits of industrial symbiosis in the forest industry. Journal of Cleaner
Production, 16(14):1536–1544, September 2008.
[10] N.B. Jacobsen. Do social factors really matter when companies engage in industrial
symbiosis? Progress in Industrial Ecology, 4(6):440–462, 2007.
[11] N.B. Jacobsen. Exploring mechanisms for mobilising industrial sustainability models
across different industrial locations. International Journal of Environmental Tech-
nology and Management, 10(3-4):442–457, 2009.
[12] David Gibbs and Pauline Deutz. Reflections on implementing industrial ecol-
ogy through eco-industrial park development. Journal of Cleaner Production,
15(17):1683–1695, November 2007.
[13] Lilian Bechara Elabras Veiga and Alessandra Magrini. Eco-industrial park develop-
ment in rio de janeiro, brazil: a tool for sustainable development. Journal of Cleaner
Production, 17(7):653–661, May 2009.
[14] A. Hewes and D.I. Lyons. The humanistic side of eco-industrial parks: Champions
and the role of trust. Regional Studies, 42(10):1329–1342, 2008.
[15] Suren Erkman and Ramesh Ramaswamy. Industrial ecological solutions. In Envi-
ronmental Solutions, pages 297–310. Academic Press, Burlington, 2005.
[16] Ewa Liwarska-Bizukojc, Marcin Bizukojc, Andrzej Marcinkowski, and Andrzej
Doniec. The conceptual model of an eco-industrial park based upon ecological rela-
tionships. Journal of Cleaner Production, 17(8):732–741, May 2009.
177
[17] C. Haskins. Multidisciplinary investigation of eco-industrial parks. Systems Engi-
neering, 9(4):313–330, 2006.
[18] Raymond P. Cote and E. Cohen-Rosenthal. Designing eco-industrial parks: a synthe-
sis of some experiences. Journal of Cleaner Production, 6(3-4):181–188, September
1998.
[19] Raymond Cote, Aaron Booth, and Bertha Louis. Eco-efficiency and SMEs in nova
scotia, canada. Journal of Cleaner Production, 14(6-7):542–550, 2006.
[20] Juliana Garcia Moretz-Sohn Monteiro, Ofelia de Queiroz Fernandes Araujo, and
Jose Luiz Medeiros. Sustainability metrics for eco-technologies assessment, part i:
preliminary screening. Clean Techn Environ Policy, 11(2):209–214, December 2008.
[21] M Aresta. Life cycle analysis applied to the assessment of the environmental impact
of alternative synthetic processes. the dimethylcarbonate case: part 1. Journal of
Cleaner Production, 7(3):181–193, March 1999.
[22] K. Krrishnamohan. Eco-industrial parks for sustainable industrial development: Ini-
tiatives in developed and developing countries. International Journal of Ecology and
Environmental Sciences, 27(3-4):137–146, 2001.
[23] Aimin Xu. Chemical Production Complex Optimization, Pollution Reduction and
Sustainable Development. PhD thesis, Louisiana State University, Louisiana State
University, December 2004.
[24] David Gibbs and Pauline Deutz. Implementing industrial ecology? planning for
eco-industrial parks in the USA. Geoforum, 36(4):452–464, July 2005.
[25] R.R. Heeres, W.J.V. Vermeulen, and F.B. de Walle. Eco-industrial park initiatives
in the USA and the netherlands: first lessons. Journal of Cleaner Production, 12(8-
10):985–995, October 2004.
178
[26] S. Leduc. Optimal location of lignocellulosic ethanol refineries with polygeneration
in sweden. Energy, 35(6):2709–2716, June 2010.
[27] M.W Melaina, O. Antonia, and M. Penev. Blending hydrogen into natural gas
pipeline networks: A review of key issues. Technical Report NREL/TP-5600-51995,
National Renewable Energy Laboratory, Golden, Colorado, March 2013.
[28] U. Bossel. Does a hydrogen economy make sense? Proceedings of the IEEE,
94(10):1826–1836, 2006.
[29] Amirhossein Hajimiragha, Michael W. Fowler, and Claudio A. Canizares. Hydrogen
economy transition in ontario canada considering the electricity grid constraints.
International Journal of Hydrogen Energy, 34(13):5275–5293, July 2009.
[30] G. J. Suppes. Roles of plug-in hybrid electric vehicles in the transition to the hydrogen
economy. International Journal of Hydrogen Energy, 31(3):353–360, March 2006.
[31] International Standards Organization. Environmental management - life cycle as-
sessment - principles and framework ISO 14040, 1997.
[32] Adisa Azapagic. Life cycle assessment and its application to process selection, design
and optimisation. Chemical Engineering Journal, 73(1):1–21, April 1999.
[33] Terry Tudor, Emma Adam, and Margaret Bates. Drivers and limitations for the
successful development and functioning of EIPs (eco-industrial parks): A literature
review. Ecological Economics, 61(2-3):199–207, March 2007.
[34] H. Gabathuler. LCA history: Centrum voor milieukunde leiden (CML) the CML
story how environmental sciences entered the debate on LCA. International Journal
of Life Cycle Assessment, 2(4):187–194, 1997.
[35] Eckhard Stormer. Greening as strategic development in industrial change - why
companies participate in eco-networks. Geoforum, 39(1):32–47, January 2008.
179
[36] Ontario Ministry of Energy and Infrastructure. Ontario electricity generation: 2005,
2005.
[37] Emmanuel D. Adamides and Yannis Mouzakitis. Industrial ecosystems as technolog-
ical niches. Journal of Cleaner Production, 17(2):172–180, January 2009.
[38] Murat Mirata and Tareq Emtairah. Industrial symbiosis networks and the contribu-
tion to environmental innovation: The case of the landskrona industrial symbiosis
programme. Journal of Cleaner Production, 13(10-11):993–1002, August 2005.
[39] Suhejla Hoti, Michael McAleer, and Laurent L. Pauwels. Modelling environmental
risk. Environmental Modelling & Software, 20(10):1289–1298, October 2005.
[40] Ivo Knoepfel. Dow jones sustainability group index: A global benchmark for corpo-
rate sustainability. Corporate Environmental Strategy, 8(1):6–15, April 2001.
[41] Ivan Kantor, Michael W. Fowler, Amirhossein Hajimiragha, and Ali Elkamel. Air
quality and environmental impacts of alternative vehicle technologies in ontario,
canada. International Journal of Hydrogen Energy, 35(10):5145–5153, May 2010.
[42] C. Yang and J. Ogden. Determining the lowest-cost hydrogen delivery mode. Inter-
national Journal of Hydrogen Energy, 32(2):268–286, 2007.
[43] Zhenhong Lin, Chien-Wei Chen, Joan Ogden, and Yueyue Fan. The least-cost hydro-
gen for southern california. International Journal of Hydrogen Energy, 33(12):3009–
3014, June 2008.
[44] Tsutomu Oi and Koichi Wada. Feasibility study on hydrogen refueling infrastructure
for fuel cell vehicles using the off-peak power in japan. International Journal of
Hydrogen Energy, 29(4):347–354, March 2004.
180
[45] Eiichi Endo. Market penetration analysis of fuel cell vehicles in japan by using
the energy system model MARKAL. International Journal of Hydrogen Energy,
32(1011):1347–1354, July 2007.
[46] Kerry-Ann Adamson. Calculating the price trajectory of adoption of fuel cell vehicles.
International Journal of Hydrogen Energy, 30(4):341–350, March 2005.
[47] J. Van Mierlo, G. Maggetto, and Ph. Lataire. Which energy source for road trans-
port in the future? a comparison of battery, hybrid and fuel cell vehicles. Energy
Conversion and Management, 47(17):2748–2760, October 2006.
[48] C. E. Thomas. Fuel cell and battery electric vehicles compared. International Journal
of Hydrogen Energy, 34(15):6005–6020, August 2009.
[49] C. E. Thomas. Comparison of transportation options in a carbon-constrained world:
Hydrogen, plug-in hybrids and biofuels, March 2008.
[50] Joan M. Ogden. Developing an infrastructure for hydrogen vehicles: a southern cali-
fornia case study. International Journal of Hydrogen Energy, 24(8):709–730, August
1999.
[51] M Campbell, D Pengally, and M Bienefeld. Air pollution burden of illness in toronto:
2004 summary, 2004.
[52] Ontario Medical Association. The illness costs of air pollution: 2005-2026 health and
economic damage estimates, 2005.
[53] S Judek, B Jessiman, D Stieb, and R Vet. Estimated number of excess deaths in
canada due to air pollution, 2005.
[54] C Koroneos, A Dompros, G Roumbas, and N Moussiopoulos. Life cycle assessment
of hydrogen fuel production processes. International Journal of Hydrogen Energy,
29(14):1443–1450, November 2004.
181
[55] Guihua Wang, Joan M. Ogden, and Daniel Sperling. Comparing air quality impacts of
hydrogen and gasoline. Transportation Research Part D: Transport and Environment,
13(7):436–448, October 2008.
[56] General Motors Corporation, BP, ExxonMobil, and Shell. Well-to-wheel energy
use and greenhouse gas emissions of advanced fuel/vehicle systems-north american
analysis; volume 3, 2001.
[57] National Hydrogen Association. The energy evolution: An analysis of alternative
vehicles and fuels to 2100, 2009.
[58] Constantine Samaras and Kyle Meisterling. Life cycle assessment of greenhouse
gas emissions from plug-in hybrid vehicles: Implications for policy. Environ. Sci.
Technol., 42(9):3170–3176, May 2008.
[59] Colin Baird and Michael McCann. Environmental Chemistry. Freeman and Com-
pany, New York, NY, 4 edition, 2008.
[60] Argonne National Laboratories. Greenhouse gases, regulated emissions and energy
use in transportation: GREET 1.8b, October 2009.
[61] A. Hajimiragha, C.A. Canizares, M.W. Fowler, and A. Elkamel. Optimal transition
to plug-in hybrid electric vehicles in ontario, canada, considering the electricity-grid
limitations. IEEE Transactions on Industrial Electronics, 57(2):690–701, February
2010.
[62] Pollution Probe. Canadians perception of advanced technology vehicles and necessary
interventions to move it to market, May 2009.
[63] Guihua Wang, Joan M. Ogden, and Michael A. Nicholas. Lifecycle impacts of natural
gas to hydrogen pathways on urban air quality. International Journal of Hydrogen
Energy, 32(14):2731–2742, September 2007.
182
[64] Dogan Keles, Martin Wietschel, Dominik Most, and Otto Rentz. Market penetration
of fuel cell vehicles analysis based on agent behaviour. International Journal of
Hydrogen Energy, 33(16):4444–4455, August 2008.
[65] Natural Resources Canada, Government of Canada. Canadian vehicle survey 2009.
Technical Report M141-18/2009, Energy Publications, Office of Energy Efficiency,
Natural Resources Canada, 2011. Includes data and analysis on transportation activ-
ities in Canada, provides estimates of fuel consumption using data on fuel purchases
and on-road vehicle use, data on number of vehicles and distance travelled.
[66] Natural Resouces Canada. Total end-use sector GHG emissions, office of en-
ergy efficiency, 2006. accessed through http://data.gc.ca/data/en/dataset/
23b85a02-a358-42b6-86c3-753b0355b616.
[67] David Pearce. The social cost of carbon and its policy implications. Oxf Rev Econ
Policy, 19(3):362–384, September 2003.
[68] L Thompson and M White. Canadian inventory of SOx, PM2.5 and PM10 emissions
resulting from the combustion of fuel oils, 2002.
[69] I. Kantor, M. Fowler, and A. Elkamel. Optimized production of hydrogen in an eco-
park network accounting for life-cycle emissions and profit. International Journal of
Hydrogen Energy, 37(6):5347–5359, 2012.
[70] S. Dunn. Hydrogen futures: Toward a sustainable energy system. International
Journal of Hydrogen Energy, 27(3):235–264, 2002.
[71] C.-J. Winter. Into the hydrogen energy economy - milestones. International Journal
of Hydrogen Energy, 30(7):681–685, 2005.
[72] A.H. Hajimiragha, C.A. Canizares, M.W. Fowler, S. Moazeni, A. Elkamel, and
S. Wong. Sustainable convergence of electricity and transport sectors in the context
183
of a hydrogen economy. International Journal of Hydrogen Energy, 36(11):6357–6375,
2011.
[73] G.F. Naterer, M. Fowler, J. Cotton, and K. Gabriel. Synergistic roles of off-peak elec-
trolysis and thermochemical production of hydrogen from nuclear energy in canada.
International Journal of Hydrogen Energy, 33(23):6849–6857, 2008.
[74] V. Utgikar and T. Thiesen. Life cycle assessment of high temperature electrolysis for
hydrogen production via nuclear energy. International Journal of Hydrogen Energy,
31(7):939–944, 2006.
[75] Jens R Rostrup-Nielsen. New aspects of syngas production and use. Catalysis Today,
63(2-4):159–164, December 2000.
[76] Linghong Zhang, Chunbao (Charles) Xu, and Pascale Champagne. Overview of
recent advances in thermo-chemical conversion of biomass. Energy Conversion and
Management, 51(5):969–982, May 2010.
[77] Jonas M Joelsson. Reduction of CO2 emission and oil dependency with biomass-
based polygeneration. Biomass and Bioenergy, 34(7):967–984, July 2010.
[78] Atul Sharma and Toshimasa Takanohashi. Controlling the H2/CO ratio of the syn-
thesis gas in a single step by catalytically gasifying coal in a steam and carbon dioxide
mixed environment at low temperatures. Energy & Fuels, 24(3):1745–1752, March
2010.
[79] A. Van Der Drift, J. Van Doorn, and J.W. Vermeulen. Ten residual biomass fuels
for circulating fluidized-bed gasification. Biomass and Bioenergy, 20(1):45–56, 2001.
[80] Xu Xiaoding. Mitigation of CO2 by chemical conversion: Plausible chemical reactions
and promising products. Energy & Fuels, 10(2):305–325, March 1996.
184
[81] A. Bandyopadhyay. Amine versus ammonia absorption of CO2 as a measure of
reducing GHG emission: A critical analysis. Clean Technologies and Environmental
Policy, 13(2):269–294, 2011.
[82] James T. Yeh, Kevin P. Resnik, Kathy Rygle, and Henry W. Pennline. Semi-batch
absorption and regeneration studies for CO2 capture by aqueous ammonia. Fuel
Processing Technology, 86(14-15):1533–1546, October 2005.
[83] Jim Andersson and Joakim Lundgren. Techno-economic analysis of ammonia pro-
duction via integrated biomass gasification. Applied Energy.
[84] George T Austin and Randolph Norris , Shreve. Shreve’s chemical process industries.
McGraw-Hill, New York, 1998.
[85] Nathan Parker, Yueyue Fan, and Joan Ogden. From waste to hydrogen: An optimal
design of energy production and distribution network. Transportation Research Part
E: Logistics and Transportation Review, 46(4):534–545, July 2010.
[86] Joan M. Ogden. Modeling infrastructure for a fossil hydrogen energy system with
CO2 sequestration. In Greenhouse Gas Control Technologies - 6th International
Conference, pages 1069–1074. Pergamon, Oxford, 2003.
[87] Environment Canada. Environment canada - national inventory report: 1990-2008.
Technical Report C2010-980262-9, Environment Canada, Ottawa, ON, Canada, De-
cember 2008.
[88] Kwi Seong Jeong and Byeong Soo Oh. Fuel economy and life-cycle cost analysis of
a fuel cell hybrid vehicle. Journal of Power Sources, 105(1):58–65, March 2002.
[89] R. Mendivil, U. Fischer, M. Hirao, and K. Hungerbuhler. A new LCA methodology
of technology evolution (TE-LCA) and its application to the production of ammonia
(1950-2000). International Journal of Life Cycle Assessment, 11(2):98–105, 2006.
185
[90] C. S Snyder. Review of greenhouse gas emissions from crop production systems
and fertilizer management effects. Agriculture, Ecosystems and Environment, 133(3-
4):247–266, October 2009.
[91] Andreas Ziegler and Michael Schroder. What determines the inclusion in a sustain-
ability stock index?: A panel data analysis for european firms. Ecological Economics,
69(4):848–856, February 2010.
[92] Carin Labuschagne, Alan C. Brent, and Ron P.G. van Erck. Assessing the sustain-
ability performances of industries. Journal of Cleaner Production, 13(4):373–385,
March 2005.
[93] Rajesh Kumar Singh, H.R. Murty, S.K. Gupta, and A.K. Dikshit. An overview of
sustainability assessment methodologies. Ecological Indicators, 9(2):189–212, March
2009.
[94] TL Saaty. The analytic hierarchy process. McGraw-Hill, New York, 1980.
[95] WL Winston. Operations research, applications and algorithms. Duxbury Press,
Belmont, CA, 3 edition, 1994.
[96] C Krotscheck. The sustainable process index a new dimension in ecological evalua-
tion. Ecological Engineering, 6(4):241–258, June 1996.
[97] Douglas Young, Richard Scharp, and Heriberto Cabezas. The waste reduction (WAR)
algorithm: environmental impacts, energy consumption, and engineering economics.
Waste Management, 20(8):605–615, December 2000.
[98] US EPA. WAR algorithm | sustainable technology | US EPA, June 2008. accessed
at http://www.epa.gov/nrmrl/std/cppb/war/sim_war.htm.
186
[99] Heriberto Cabezas, Jane C. Bare, and Subir K. Mallick. Pollution prevention with
chemical process simulators: the generalized waste reduction (WAR) algorithmfull
version. Computers & Chemical Engineering, 23(4-5):623–634, May 1999.
[100] G. Siracusa, A. D. La Rosa, and S. E. Sterlini. A new methodology to calculate
the environmental protection index (ep). a case study applied to a company pro-
ducing composite materials. Journal of Environmental Management, 73(4):275–284,
December 2004.
[101] Khandoker A. Hossain, Faisal I. Khan, and Kelly Hawboldt. E-green a robust risk-
based environmental assessment tool for process industries. Ind. Eng. Chem. Res.,
46(25):8787–8795, December 2007.
[102] X. Dongfeng, Y. Li, S. Jingzhu, and S. Hu. Synthesis of eco-industrial system con-
sidering environmental value using adaptive simulated annealing genetic algorithms,
volume 15. 2003.
[103] Lingmei Wang, Jintun Zhang, and Weidou Ni. Emergy evaluation of eco-industrial
park with power plant. Ecological Modelling, 189(1-2):233–240, November 2005.
[104] Lingmei Wang, Weidou Ni, and Zheng Li. Emergy evaluation of combined heat
and power plant eco-industrial park (CHP plant EIP). Resources, Conservation and
Recycling, 48(1):56–70, July 2006.
[105] Jin-su Kang, Hyeong-dong Kim, and Tai-yong Lee. Sharing benefits of eco-industrial
park by multiobjective material flow optimization. In Process Systems Engineering
2003, 8th International Symposium on Process Systems Engineering, volume Volume
15, Part 1, pages 499–504. Elsevier, 2003.
[106] Jan J. Lerou and Ka M. Ng. Chemical reaction engineering: A multiscale approach to
a multiobjective task. Chemical Engineering Science, 51(10):1595–1614, May 1996.
187
[107] Environment Canada. Environment canada - air - pollutants, 2011. acessed at
http://www.ec.gc.ca/air/default.asp?lang=En&n=BCC0B44A-1.
[108] Ivan Kantor, Ali Elkamel, and Michael W. Fowler. Optimization of material and
energy exchange in an eco-park network considering three fuel sources. International
Journal of Advanced Operations Management, In Press, 2014.
[109] Society of Automotive Engineers. The Automotive Fuel Cell Technology Report: 2013
Edition. SupplierBusiness, 2013 edition, April 2013.
[110] Kok Siew Ng, Nan Zhang, and Jhuma Sadhukhan. Techno-economic analysis of
polygeneration systems with carbon capture and storage and CO2 reuse. Chemical
Engineering Journal, 219:96–108, March 2013.
[111] L. Sokka, S. Lehtoranta, A. Nissinen, and M. Melanen. Analyzing the environmental
benefits of industrial symbiosis: Life cycle assessment applied to a finnish forest
industry complex. Journal of Industrial Ecology, 15(1):137–155, 2011.
[112] Tuomas Mattila, Suvi Lehtoranta, Laura Sokka, Matti Melanen, and Ari Nissinen.
Methodological aspects of applying life cycle assessment to industrial symbioses.
Journal of Industrial Ecology, 16(1):5160, 2012.
[113] Bo Pedersen Weidema and Marianne Suhr Wesnæs. Data quality management for
life cycle inventoriesan example of using data quality indicators. Journal of Cleaner
Production, 4(34):167–174, 1996.
[114] Ernest A. Lowe and Laurence K. Evans. Industrial ecology and industrial ecosystems.
Journal of Cleaner Production, 3(1-2):47–53, 1995.
[115] Steven J Smith, Hugh Pitcher, and T.M.L Wigley. Global and regional anthropogenic
sulfur dioxide emissions. Global and Planetary Change, 29(12):99–119, May 2001.
188
[116] A Shafeen, E Croiset, P. L Douglas, and I Chatzis. CO2 sequestration in ontario,
canada. part i: storage evaluation of potential reservoirs. Energy Conversion and
Management, 45(17):2645–2659, October 2004.
[117] Emil Raymond Riegel and James Albert Kent. Riegel’s handbook of industrial chem-
istry. Van Nostrand Reinhold, 1992.
[118] I. Y. Kim and O. L. de Weck. Adaptive weighted-sum method for bi-objective opti-
mization: Pareto front generation. Struct Multidisc Optim, 29(2):149–158, February
2005.
[119] Eva M. Lovelady and Mahmoud M. El-Halwagi. Design and integration of eco-
industrial parks for managing water resources. Environ. Prog. Sustainable Energy,
28(2):265–272, July 2009.
[120] John Ehrenfeld and Nicholas Gertler. Industrial ecology in practice: The evolution of
interdependence at kalundborg. Journal of Industrial Ecology, 1(1):67–79, January
1997.
[121] K Lee. Establishment of eco-industrial parks for infrastructure of cleaner production.
In Presentation Manuals at the Korea National Cleaner Production Center, South
Korea, South Korea, 2005.
[122] Hung-Suck Park, Eldon R. Rene, Soo-Mi Choi, and Anthony S. F. Chiu. Strategies for
sustainable development of industrial park in ulsan, south KoreaFrom spontaneous
evolution to systematic expansion of industrial symbiosis. Journal of Environmental
Management, 87(1):1–13, April 2008.
[123] Song Hwa Chae. Optimization of a waste heat utilization network in an eco-industrial
park. Applied Energy, 87(6):1978–1988, June 2010.
189
[124] I. Fernandez and M. C. Ruiz. Descriptive model and evaluation system to locate
sustainable industrial areas. Journal of Cleaner Production, 17(1):87–100, January
2009.
[125] Cristina Sendra, Xavier Gabarrell, and Teresa Vicent. Material flow analysis adapted
to an industrial area. Journal of Cleaner Production, 15(17):1706–1715, November
2007.
[126] Yan Zhao, Jin-cheng Shang, Chong Chen, and He-nan Wu. Simulation and evaluation
on the eco-industrial system of changchun economic and technological development
zone, china. Environ Monit Assess, 139(1-3):339–349, April 2008.
[127] D Spriggs, E Lowe, J Watz, and Mahmoud M. El-Halwagi. Design and development of
eco-industrial parks. 2004 AIChE Spring National Meeting, Conference Proceedings;
New Orleans, LA; United States; 25 April 2004 through 29 April 2004, pages 2420–
2430, April 2004.
[128] James Scott Baldwin. Industrial ecosystems: an evolutionary classification scheme.
Progress in Industrial Ecology, An International Journal, 5(4):277, 2008.
[129] Goran Finnveden, Michael Z. Hauschild, Tomas Ekvall, Jeroen Guinee, Reinout Hei-
jungs, Stefanie Hellweg, Annette Koehler, David Pennington, and Sangwon Suh. Re-
cent developments in life cycle assessment. Journal of Environmental Management,
91(1):1–21, October 2009.
[130] Reinout Heijungs, Gjalt Huppes, and Jeroen B. Guinee. Life cycle assessment and
sustainability analysis of products, materials and technologies. toward a scientific
framework for sustainability life cycle analysis. Polymer Degradation and Stability,
95(3):422–428, March 2010.
190
[131] Calin-Cristian Cormos. Assessment of hydrogen and electricity co-production
schemes based on gasification process with carbon capture and storage. Interna-
tional Journal of Hydrogen Energy, 34(15):6065–6077, August 2009.
[132] J. Francois, L. Abdelouahed, G. Mauviel, F. Patisson, O. Mirgaux, C. Rogaume,
Y. Rogaume, M. Feidt, and A. Dufour. Detailed process modeling of a wood gasi-
fication combined heat and power plant. Biomass and Bioenergy, 51:68–82, April
2013.
[133] C. Koroneos, A. Dompros, and G. Roumbas. Hydrogen production via biomass
gasificationA life cycle assessment approach. Chemical Engineering and Processing:
Process Intensification, 47(8):1261–1268, August 2008.
[134] Siyu Yang, Qingchun Yang, Hengchong Li, Xing Jin, Xiuxi Li, and Yu Qian.
An integrated framework for modeling, synthesis, analysis, and optimization of
coal gasification-based energy and chemical processes. Ind. Eng. Chem. Res.,
51(48):15763–15777, December 2012.
[135] Dustin Jones, Debangsu Bhattacharyya, Richard Turton, and Stephen E. Zitney.
Optimal design and integration of an air separation unit (ASU) for an integrated
gasification combined cycle (IGCC) power plant with CO2 capture. Fuel Processing
Technology, 92(9):1685–1695, September 2011.
[136] M. K Mann and P. L Spath. Life cycle assessment of a biomass gasification combined-
cycle power system. Technical Report NREL/TP–430-23076, OSTI ID: 567454, Na-
tional Renewable Energy Lab., Golden, CO (US), December 1997.
[137] Philip Nuss, Kevin H. Gardner, and Jenna R. Jambeck. Comparative life cycle as-
sessment (LCA) of construction and demolition (C&D) derived biomass and U.S.
northeast forest residuals gasification for electricity production. Environ. Sci. Tech-
nol., 47(7):3463–3471, April 2013.
191
[138] PRe Sustainability. SimaPro LCA Software, January 2011.
[139] Kay Damen, Martijn van Troost, Andre Faaij, and Wim Turkenburg. A comparison
of electricity and hydrogen production systems with CO2 capture and storage. part
a: Review and selection of promising conversion and capture technologies. Progress
in Energy and Combustion Science, 32(2):215–246, 2006.
[140] Mar Perez-Fortes, Pol Arranz-Piera, Jose Miguel Laınez, Enric Velo, and Luis Puig-
janer. Optimal location of gasification plants for electricity production in rural areas.
In 21st European Symposium on Computer Aided Process Engineering, volume Vol-
ume 29, pages 1809–1813. Elsevier, 2011.
[141] R. Strube, G. Pellegrini, and G. Manfrida. The environmental impact of post-
combustion CO2 capture with MEA, with aqueous ammonia, and with an aqueous
ammonia-ethanol mixture for a coal-fired power plant. Energy, 36(6):3763–3770,
June 2011.
[142] Jeremy David and Howard Herzog. The cost of carbon capture. In Proceedings of 5th
International Conference on Greenhouse Gas Control Technologies, pages 985–990,
Cairns,Australia, 2001. CSIRO.
[143] Umberto Desideri, Livia Arcioni, and Marco Tozzi. Feasibility study for a carbon cap-
ture and storage project in northern italy. International Journal of Energy Research,
32(12):1175–1183, 2008.
[144] Jose D. Figueroa, Timothy Fout, Sean Plasynski, Howard McIlvried, and Ramesh-
war D. Srivastava. Advances in CO2 capture technologyThe U.S. department of
energy’s carbon sequestration program. International Journal of Greenhouse Gas
Control, 2(1):9–20, January 2008.
192
[145] Elizabeth Lokey. Valuation of carbon capture and sequestration under greenhouse
gas regulations. The Electricity Journal, 22(4):11–24, May 2009.
[146] Kok Siew Ng and Jhuma Sadhukhan. Techno-economic performance analysis of
bio-oil based fischer-tropsch and CHP synthesis platform. Biomass and Bioenergy,
35(7):3218–3234, July 2011.
[147] D Singh, E Croiset, P.L Douglas, and M.A Douglas. Techno-economic study of CO2
capture from an existing coal-fired power plant: MEA scrubbing vs. O2/CO2 recy-
cle combustion. Energy Conversion and Management, 44(19):3073–3091, November
2003.
[148] Soren Anderson and Richard Newell. Prospects for carbon capture and storage tech-
nologies. Annual Review of Environment and Resources, 29:109–142, 2004.
[149] Diego Iribarren, Ana Susmozas, and Javier Dufour. Life-cycle assessment of Fis-
cherTropsch products from biosyngas. Renewable Energy, 59:229–236, November
2013.
[150] K. J Li and Georgios Protopapas. CO2 capture for sequestration. International
Petroleum Technology Conference 2007, 3:var.pagings, 2007.
[151] Naser A. Odeh and Timothy T. Cockerill. Life cycle analysis of UK coal fired power
plants. Energy Conversion and Management, 49(2):212–220, February 2008.
[152] Gianluca Valenti, Davide Bonalumi, and Ennio Macchi. Energy and exergy analyses
for the carbon capture with the chilled ammonia process (CAP). Energy Procedia,
1(1):1059–1066, 2009.
[153] Qun Yi, Yang Fan, Wenying Li, and Jie Feng. CO2 capture and use in a novel coal-
based polygeneration system. Ind. Eng. Chem. Res., 52(39):14231–14240, September
2013.
193
[154] Minh T. Ho, Guy W. Allinson, and Dianne E. Wiley. Reducing the cost of CO2
capture from flue gases using pressure swing adsorption. Ind. Eng. Chem. Res.,
47(14):4883–4890, July 2008.
[155] Oliver J. Smith IV and Arthur W. Westerberg. The optimal design of pressure swing
adsorption systems. Chemical Engineering Science, 46(12):2967–2976, 1991.
[156] S. Sircar and T. C. Golden. Purification of hydrogen by pressure swing adsorption.
Separation Science and Technology, 35(5):667–687, 2000.
[157] Hsien H. Khoo and Reginald B. H. Tan. Life cycle investigation of CO2 recovery and
sequestration. Environ. Sci. Technol., 40(12):4016–4024, June 2006.
[158] Jeremy David. Economic evaluation of leading technology options for sequestration of
carbon dioxide. Thesis, Massachusetts Institute of Technology, 2000. Thesis (S.M.)–
Massachusetts Institute of Technology, Technology and Policy Program, 2000.
[159] H. Herzog, D. Golomb, and S. Zemba. Feasibility, modeling and economics of se-
questering power plant CO2 emissions in the deep ocean. Environmental Progress,
10(1):6474, 1991.
[160] Henriette Undrum, Olav Bolland, and Eivind Aarebrot. Economical assessment of
natural gas fired combined cycle power plant with CO2 capture and sequestration.
In Proceedings of the Fifth International Conference on Greenhouse Gas Control
Technologies, Cairns, Australia, August 2000.
[161] Fred Kozak, Arlyn Petig, Ed Morris, Richard Rhudy, and David Thimsen. Chilled
ammonia process for CO2 capture. Energy Procedia, 1(1):1419–1426, 2009.
[162] W. Zhang and E.-K. An. Life cycle analysis of coal-based ammonia fuel for vehicular
application. Meitan Xuebao/Journal of the China Coal Society, 35(1):173–176, 2010.
194
[163] Jorge Cristobal, Gonzalo Guillen-Gosalbez, Laureano Jimenez, and Angel Irabien.
Multi-objective optimization of the electricity production from coal burning. In 21st
European Symposium on Computer Aided Process Engineering, volume Volume 29,
pages 1814–1818. Elsevier, 2011.
[164] Nagore Sabio, Mamdouh Gadalla, Laureano Jimenez, and Gonzalo Guillen-Gosalbez.
Multi-objective optimization of a hydrogen supply chain for vehicle use including eco-
nomic and financial risk metrics. a case study of spain. In 20th European Symposium
on Computer Aided Process Engineering, volume Volume 28, pages 121–126. Elsevier,
2010.
[165] E. Arzate, P. Huitzil, A. Gonzalez, B.E. Martınez, and I.E. Grossmann. Automated
optimization model to perform sensitivity analysis on cost of investment required
to upgrade treatment plants in water networks. In Ian David Lockhart Bogle and
Michael Fairweather, editor, Computer Aided Chemical Engineering, volume Volume
30 of 22nd European Symposium on Computer Aided Process Engineering, pages
1063–1067. Elsevier, 2012.
[166] F. Hernandez-Sancho, M. Molinos-Senante, and R. Sala-Garrido. Cost modelling for
wastewater treatment processes. Desalination, 268(13):1–5, March 2011.
[167] Elorri Igos, Alice Dalle, Ligia Tiruta-Barna, Enrico Benetto, Isabelle Baudin, and
Yoann Mery. Life cycle assessment of water treatment: what is the contribution of
infrastructure and operation at unit process level? Journal of Cleaner Production,
65:424–431, February 2014.
[168] Eusiel Rubio-Castro, Jose Marıa Ponce-Ortega, Medardo Serna-Gonzalez, and Mah-
moud M. El-Halwagi. Optimal reconfiguration of multi-plant water networks into an
eco-industrial park. Computers & Chemical Engineering, 44:58–83, September 2012.
195
[169] Zengwei Yuan, Weili Jiang, and Jun Bi. Cost-effectiveness of two operational mod-
els at industrial wastewater treatment plants in china: A case study in shengze
town, suzhou city. Journal of Environmental Management, 91(10):2038–2044, Octo-
ber 2010.
[170] Elorri Igos, Enrico Benetto, Isabelle Baudin, Ligia Tiruta-Barna, Yoann Mery, and
Damien Arbault. Cost-performance indicator for comparative environmental assess-
ment of water treatment plants. Science of The Total Environment, 443:367–374,
January 2013.
[171] A. Racoviceanu, B. Karney, C. Kennedy, and A. Colombo. Life-cycle energy use
and greenhouse gas emissions inventory for water treatment systems. Journal of
Infrastructure Systems, 13(4):261–270, 2007.
[172] Helene Slagstad and Helge Brattebø. Life cycle assessment of the water and wastew-
ater system in trondheim, norway a case study. Urban Water Journal, 0(0):1–12,
2013.
[173] Le Tong, Xin Liu, Xuewei Liu, Zengwei Yuan, and Qiong Zhang. Life cycle assessment
of water reuse systems in an industrial park. Journal of Environmental Management,
129:471–478, November 2013.
[174] Nathan Parker. Using natural gas transmission pipeline costs to estimate hydro-
gen pipeline costs. Institute of Transportation Studies, University of California,
Davis, Research Report UCD-ITS-RR-04-35, University of California, Davis, De-
cember 2004.
[175] Zhenhua Rui, Paul A. Metz, Douglas B. Reynolds, Gang Chen, and Xiyu Zhou.
Regression models estimate pipeline construction costs. Oil & Gas Journal,
109(14):120–127, July 2011.
196
[176] Transport Canada. Estimation of unit costs of rail transportation in canada final
report. Technical Report T22-149/2007-1E-PDF, DAMF Consultants Inc. in associ-
ation with Joseph Schulman Consulting and CANARAIL Consultants Inc., Canada,
2007.
[177] Mike Kennedy, Rich Wong, Andrew Vandenbroek, Dave Lovekin, and Marlo
Raynolds. Biomass sustainability analysis. FINAL REPORT, revision c, Pembina
Institute, April 2011.
[178] Laura A. Pellegrini, Giorgio Soave, Simone Gamba, and Stefano Lange. Eco-
nomic analysis of a combined energymethanol production plant. Applied Energy,
88(12):4891–4897, December 2011.
[179] Mehdi Aghaei Meybodi and Masud Behnia. Optimum sizing of the prime mover in a
medium scale gas turbine CHP system. J. Eng. Gas Turbines Power, 133(11):112001–
112001, May 2011.
[180] Qin Chen, Ashok Rao, and Scott Samuelsen. H2 coproduction in IGCC with CCS via
coal and biomass mixture using advanced technologies. Applied Energy, 118:258–270,
April 2014.
[181] Jeffrey Ralph Bartels. A feasibility study of implementing an Ammonia Economy.
Master’s thesis, Iowa State University, Iowa State University, 2008.
[182] A. K. Roy. Why not coal-based fertiliser plants? Economic and Political Weekly,
May 2001.
[183] Robert C. Rosaler. Standard Handbook of Plant Engineering, Third Edition.
McGRAW-HILL: New York, Chicago, San Francisco, Lisbon, London, Madrid, Mex-
ico City, Milan, New Delhi, San Juan, Seoul, Singapore, Sydney, Toronto, 3 edition,
2002.
197
Appendix: Additional detail with
respect to primary modeling in
Chapter 3
The following is a summary of the AMPL model which was primarily constructed by
Amirhossein Hajimiragha as presented in [29, 61] which yielded the supportable penetra-
tion of alternative fuel vehicles used in Chapter 3. These estimates were then used to
calculate the emission reductions from these vehicles using GREET 1.8b as described in
Chapter 3. A summary of the model, as published by Hajimiragha et al. is shown below
with permission from the International Association of Hydrogen Energy.
Ω =λiπi
=λ∑
i bi∑i πibi
Describes the ratio of base-load growth rate (λi) to peak-load growth rate (πi) under the
assumption that this ratio is constant across all zones. Thus λ is the annual baseload
growth rate in all zones and bi is the base load value in Zone i. The growth rate in each
zone can then be calculated:
λi = λπi∑
i bi∑i πibi
198
Phppiy is the total installed HPP capacity in Zone i, by Year y which can by utilized
for producing hydrogen locally and also in other zones. This total capacity is composed of
smaller power components, Psijy, which represents the contribution of zone i to the total
power required in zone j to be transported by compressed gas truck transportation. Other
zones can share in part of the required power in zone i based on the complementary power
component Psjiy. Accordingly, the required power of HPPs in Zone i by Year y (Phiy) can
be expressed as follows:
Phiy = Phppiy −∑j 6=i
Psijy +∑j 6=i
Psjiy ∀i, j ∈ Z ∧ y ∈ Υ1
Phppiy = Phppiy−1 + ∆Phppiy
where:
• ∆Phppiy is the newly installed HPP in Zone i and Year y;
• Z = 1,...,10 is the set of indices of zones or buses in the simplified network; and,
• Υ1 = 2009,...,2025 is the set of indices of planning years starting in 2009.
Operational hours or capacity factors for the HPPs must be considered to link the
power component to the amount of hydrogen transferred between zones. The HHV of
hydrogen coupled with assumptions of 68 hours of operation per week with 70 % plant
efficiency yields the power component, Psijy, is capable of producing 0.1724 Psijy tonnes
of hydrogen per day for transfer to zone j by compressed gas tube trailers. This transfer is
represented by Thijy.
199
Thijy = 0.1724Psijy ∀(i, j) ∈ Z∗ ∧ y ∈ Υ
where Z∗ = (i, j) : i, j ∈ Z, i 6= j is the set of indices of hydrogen transfer corridors
and Υ = 2008, ..., 2025 is the set of indices of planning years starting from 2008.
The total cost of transporting hydrogen between zone i and j, TOCijy , follows a step
function depending on the total transfer of hydrogen:
TOCijy = m ·OCy · dij
where m stands for the number of trucks, OCy is operational cost in year y per com-
pressed gas truck [$CADkm−1] and dij is the distance between zones i and j in km.
The objective function in this optimization is to minimize costs for electricity and hy-
drogen transportation. The cost function is composed of the costs and revenues from
electricity import/export in addition to generation costs for segments of 8 weekday hours
(0:00 - 07:00) and 14 weekend hours (0:00-13:00) with an additional consideration of hy-
drogen tranportation costs.
∑y∈Υ1
∑i∈Z
(Pgω1
iy + Pimω1iy − Pex
ω1iy
)·HOEP ω1
y × 8× 261
(PGω2
iy + Pimω2iy − Pex
ω2iy
)·HOEP ω2
y × 14× 104
+∑y∈Υ1
∑(i,j)∈Z∗
(ntry∑m=1
m ·Kmijy
)2OCy · dij × 365 +
DR · CCcab[1− (1 +DR)−LTcab ]
+DR · CCtube
[1− (1 +DR)−LTtube ]
200
where:
• ω1 is an index for the time period corresponding to 8 weekday hours (12 am-7 am);
• ω2 is an index for the time period corresponding to 14 weekend hours (12 am-1 pm);
• Pg, Pim and Pex are zonal generation power, imported power, and exported power,
respectively, in Zone i, Year y, and during the time periodω1 orω2;
• Kmijy is a binary variable which takes the value of 1 if m trucks are needed for daily
hydrogen transfer between Zones i and j in Year y;
• CCcab and CCtube are the capital cost of cab and tube trailers, respectively;
• LTcab and LTtube are the life time of cab and tube trailers, respectively;
• DR is the discount rate; and,
• ntry is the maximum number of compressed gas trucks in route between Zones i and
j in Year y.
The power losses in line (i,j) of the electricity network can be approximately calculated
as:
Plossij ∼= gij (δi − δj)2
where gij is the conductance of the line between buses i and j, and δ denotes the cor-
responding bus voltage angles according to the following equation:
δγijy = |δγiy − δγjy|
201
Which can be approximated in linear terms by the following equations:
δγijy = δγ+ijy + δγ−ijy
δγiy − δγjy = δγ+
ijy + δγ−ijy
δγ+ijy ≥ 0
δγ−ijy ≥ 0
∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ
where Ω is the set of indices of transmission lines and Ψ = ω1, ω2. A linear approx-
imation of power losses in Year y and during the time period γ can be obtained using L
piecewise linear blocks as follows:
δγijy =L∑l=1
δγijy(l)
Plossγijy = gijy
L∑l=1
αijy(l)δγijy(l)
where αijy(l) and δγijy(l) represent the slope and value of the lth block of voltage angle,
respectively. Assuming that each angle block has a constant length ∆δy, the slope of the
blocks of angles for all lines (i,j) can be calculated as:
αijy(l) = (2l − 1)∆δy ∀(i, j) ∈ Ω ∧ y ∈ Υ
Enforcing the adjacency of the angle blocks requires the following:
202
δγijy(l) ≥ 0 ∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ ∧ l ∈ L1
ωγijy(l) ·∆δy ≤ δγijy(l) ∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ ∧ l ∈ L2
δγijy(l) ≤ ωγijy(l − 1) ·∆δy ∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ ∧ l ∈ L3
ωγijy(l) ≤ ωγijy(l − 1) ∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ ∧ l ∈ L4
where ωγijy(l) is a binary variable which takes the value of 1 if the value of the lth angle
block for the line (i,j) is equal to its maximum value ∆δy; L1 = 1, . . . , L; L2 = 1, . . . , L− 1;
L3 = 2, . . . , L; and L4 = 2, . . . , L− 1.
Considering the line losses model just described, the net power injected at Zone i can be
represented as:
Piy =∑
(i,j)∈Ω
[1
2gijy
L∑l=1
αijy(l)δγijy(l)− bijy(δ
γiy − δ
γjy)
]
where bijy is the susceptance of the line (i,j) in Year y. Consequently, in general terms,
the zonal power balance constraints can be formulated as follows:
Pgγiy − Plγiy + Pimγ
iy − Pexγiy −
∑(i,j)∈Ω
[1
2gijy
L∑l=1
αijy(l)δγijy(l)− bijy(δ
γiy − δ
γjy)
]= 0
∀i ∈ Z ∧ y ∈ Υ ∧ γ ∈ Ψ
where Pl is the total load in each zone and is comprised of zonal electricity demand
(Pe) and total installed HPPs as follows:
Plγiy − Phppiy − Peγiy = 0 ∀i ∈ Z ∧ y ∈ Υ ∧ γ ∈ Ψ
The power generation in each year and zone is bounded by minimum and maximum
limits Pgiy
and Pgiy, respectively. These limits are the minimum and maximum effective
203
generation capacities which are available in each zone during the planning years, resulting
in the following inequality constraint:
Pgiy≤ Pgγiy ≤ Pgiy∀i ∈ Z ∧ y ∈ Υ ∧ γ ∈ Ψ
These limits are stated as:
Pimiy ≤ Pimγiy ≤ Pimiy ∀i ∈ Z ∧ y ∈ Υ ∧ γ ∈ Ψ
Pexiy ≤ Pexγiy ≤ Pexiy ∀i ∈ Z ∧ y ∈ Υ ∧ γ ∈ Ψ
where Pimiy and Pimiy are lower and upper bounds of imported power, respectively; addi-
tionally, Pexiy and Pexiy are exported power minimum and maximum limits, respectively.
These constraints are defined as:
bij(δγiy − δ
γjy) +
1
2gij
L∑l=1
αijy(l)δγijy(l) ≤ Pdij − bij(δγiy − δ
γjy)
+1
2gij
L∑l=1
αijy(l)δγijy(l) ≤ Prij
∀(i, j) ∈ Ω ∧ y ∈ Υ ∧ γ ∈ Ψ
where Pdij and Prij are maximum capacity of the transmission corridor (i,j) in direct and
reverse power flow, respectively.
In order to effectively model the hydrogen transportation costs, the following constraints
204
are also needed:
[C(m− 1) + ε]Kmijy ≤ thmijy ≤ C ·m ·Kmijy
ntry∑m=1
Kmijy ≤ 1
Thijy =
ntry∑m=1
thmijy
∀m ∈M ∧ (i, j) ∈ Z∗ ∧ y ∈ Υ1
where ε is a very small positive number; thmijy is an auxiliary variable representing the
transferred hydrogen, since Thijy = thmijy if Kmijy = 1; and M = 1, . . . , ntry.
These limits are represented by:
0 ≤ Phppiy ≤ Phppiy ∀i ∈ Z ∧ y ∈ Υ
where Phppiy is the maximum size of HPP which is allowed to be installed in zone i by
Year y. Since HPP is not installed for 2008, Phppiy is equal to zero for y = 2008.
205